Moves create_session, list_sessions, send_to_session and manage_session out of
ai_interaction.py into src/agent_tools/session_tools.py (the do_ prefix
dropped) and registers them in TOOL_HANDLERS, so dispatch flows through the
registry instead of the dispatch_ai_tool elif in tool_execution.py. Same
pattern as the model-interaction move.
The bodies move verbatim; each fetches the runtime-set session manager via a
get_session_manager() shim, and reuses _resolve_model / AI_CHAT_TIMEOUT from
ai_interaction. manage_session's internal 'list' alias is repointed from the
old do_list_sessions to the moved list_sessions. stream_ai_tool (dead, no
callers) and do_pipeline stay put. dispatch_ai_tool loses its four now-unused
branches.
Tests: test_session_tools_registry covers registration, owner threading, the
manage_session->list_sessions delegation, graceful no-manager handling, and
registry dispatch. Verified end-to-end against a live SessionManager.
Detached bash jobs (#!bg) could be launched and auto-reported on completion,
but the agent had no way to act on a running one: no on-demand output read and
no kill (it blocked until the 1h max-runtime). bg_jobs had the pieces
(_read_output, list_for_session, internal _kill) but none was exposed.
Adds:
- bg_jobs.kill(job_id): tears down the process tree, marks the job killed, and
sets followed_up so the monitor does not also auto-continue a deliberate kill.
- manage_bg_jobs registry tool with actions list / output / kill, scoped to the
chat that launched the job (cross-session access reads as not found).
- Wiring: TOOL_HANDLERS/TAGS, function schema, RAG index + keyword hints, parser
name map, dispatch (threads session_id via _direct_fallback). Gated like bash
(NON_ADMIN_BLOCKED_TOOLS; plan-mode mutator).
- agent_loop: background-job intent regex maps to the files domain (and the tool
joins _DOMAIN_TOOL_MAP[files]) so short commands like 'kill that job' are not
dropped by the low-signal gate that skips tool retrieval.
- bg launch message tells the model to call manage_bg_jobs itself for check/stop
rather than printing raw tool syntax to the user.
Tests: tests/test_bg_job_tools.py (kill semantics, per-chat scoping, actions,
and the intent classifier).
* fix(tools): prune skipped dirs before descending in glob tool
GlobTool used pathlib.Path.rglob which descends into every directory
(including node_modules, .git, dist, etc.) and filters AFTER the walk.
On repos with large junk directories this causes the glob tool to hang
for minutes.
Replace rglob with os.walk that prunes _CODENAV_SKIP_DIRS before
descending — matching the approach GrepTool already uses. Also add a
fast path for literal patterns (no wildcards → direct path lookup).
Fixes#4493
* fix(tools): use regex glob matching to fix * semantics and literal fallback
Replace fnmatch with _glob_to_regex so that * stays within a single
path segment (matching pathlib/rglob semantics) and **/ spans zero or
more directories. Literal patterns now fall through to os.walk when
the direct path lookup misses, so e.g. 'foo.py' still finds files at
any depth.
Add tests for:
- bare literal matching in subdirectories
- multi-segment single-star patterns (sub/*.txt)
- * not crossing / boundaries
- ** matching at arbitrary depth
Closes#4493
---------
Co-authored-by: michaelxer <michaelxer@users.noreply.github.com>
The harmony stream router only recognized the analysis and final channels, so
gpt-oss's standard `commentary` channel (tool-call preambles / function-arg
bodies) was unhandled: the literal `<|channel|>commentary` marker, the
`to=functions.*` recipient, and the commentary body all leaked into the
visible answer. Add commentary to the marker regex + the suffix-hold table, and
route its body to thinking (only `final` is user-facing). Adds a regression
test (split-chunk + recipient + body), verified to fail without the fix.
* fix(agent): index api_call so RAG tool selection can retrieve it
api_call exists in FUNCTION_TOOL_SCHEMAS and the agent's system prompt
advertises configured API integrations, but the tool had no entry in
BUILTIN_TOOL_DESCRIPTIONS. RAG tool selection embeds those descriptions and
retrieves the top-K per message, so a tool without one can never be selected:
the agent claims it can call Home Assistant/Miniflux/Gitea/etc. and then
never receives the api_call schema (unless the Personal Assistant
ASSISTANT_ALWAYS_AVAILABLE path applies).
Add a retrieval-rich description for api_call, plus an ast-based parity test
asserting every FUNCTION_TOOL_SCHEMAS tool has an index description so the
next added tool cannot silently drift the same way.
Fixes#3794
* fix(agent): route API-integration intent to api_call at selection time
Addresses review (RaresKeY) on #3923: indexing api_call in the ToolIndex
description was necessary but not sufficient — the #3794 repro ('Use the
api_call tool to call Home Assistant GET /api/states') matched no domain in
_classify_agent_request, classified as low-signal, so the agent loop skipped
retrieval entirely and the schema filter sent only ALWAYS_AVAILABLE
(manage_memory/ask_user/update_plan). api_call never reached the model.
- _classify_agent_request: detect API-integration intent (api_call,
integration(s), Home Assistant/Miniflux/Gitea/Linkding/Jellyfin) -> new
'integrations' domain, so the turn is no longer low-signal.
- _DOMAIN_TOOL_MAP['integrations'] = {api_call}: deterministically seeds
api_call into relevant tools after retrieval, independent of embeddings.
- _DOMAIN_RULES['integrations']: rule pack (required — _domain_rules_for_tools
indexes _DOMAIN_RULES[domain] directly).
- tool_index _KEYWORD_HINTS: parity hint for the retrieval / keyword-fallback
paths.
- Regression drives the real classifier -> domain-map -> FUNCTION_TOOL_SCHEMAS
filter chain and asserts api_call is advertised for the #3794 prompt.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Moves chat_with_model, ask_teacher and list_models out of ai_interaction.py
into src/agent_tools/model_interaction_tools.py (the do_ prefix dropped) and
registers them in TOOL_HANDLERS, so dispatch flows through the registry instead
of the dispatch_ai_tool elif in tool_execution.py.
The implementations are relocated, not wrapped. ai_interaction.py keeps only
the shared helpers they reuse (_resolve_model, AI_CHAT_TIMEOUT), still used by
the not-yet-migrated session/pipeline tools. dispatch_ai_tool loses its three
now-unused branches.
Also removes the dead do_second_opinion: it was already off the live tool
surface (no tag/schema/parsing/dispatch; tool_index.py notes it was removed),
so the function and its stale frontend catalog entries (admin.js, assistant.js)
are deleted.
Tests: owner-scope test points at the new list_models location and drops the
moved tools from the dispatch_ai_tool parametrize; a new
test_model_interaction_registry covers registration, owner threading, and
registry dispatch.
* fix(security): allowlist manage_mcp 'add' to close the agent-path RCE
do_manage_mcp('add') passed model- and prompt-injection-controlled command,
args, and env straight to a stdio subprocess spawn with no validation, and it
persisted an enabled server row before connecting (so a payload also survived
to re-execute on restart). A string smuggled into a skill description, memory
entry, fetched page, or email body could register a server running arbitrary
code as the app UID, e.g. command='sh' args=['-c','...'].
Add _validate_mcp_command, applied on the agent path before any DB write or
spawn:
- Hard-deny interpreters, runtimes, package runners, shells, and exec-wrappers
(even if an operator lists one in ODYSSEUS_MCP_ALLOWED_COMMANDS).
- Require a bare basename (no path components, no shell metacharacters) that is
present in the operator allowlist (empty by default).
- Reject code-exec argv flags by prefix so glued forms are caught too
(-c/-e/-m/--eval/--exec/--print/--module/--command/--require), remote-URL
args, and env keys that inject code into the child (LD_PRELOAD, NODE_OPTIONS,
PYTHONPATH, DYLD_*, PATH, ...).
A rejected registration returns an error, writes no row, and makes no
connection. The trusted admin route is unchanged. Mirrors the policy intent of
_validate_serve_cmd but inverted for the model-reachable surface.
Supersedes #438; incorporates the bypass forms found in its review (interpreter
script paths, -m pip, glued -c/-e, --eval=, eval subcommands, package runners,
remote URLs) and adds integration coverage on the real do_manage_mcp path.
Closes#2891
* fix(security): deny versioned/alias runtimes in manage_mcp allowlist
Addresses RaresKeY's review on #4433. The hard-deny matched command names
exactly, so versioned or alias runtime forms (python3.11, node18, pip3,
ruby3.2, java, javac, bunx, tsx, ts-node, pypy3, ...) slipped past and, if an
operator allowlisted one, re-opened the prompt-injection-controlled MCP
registration path.
- Canonicalize a trailing version suffix before the deny check so versioned
forms collapse to the family (python3.11 -> python, node18 -> node, pip3 ->
pip); both the raw basename and the canonical form are denied.
- Broaden the denied-family set (java/javac/jshell/jbang/kotlin/dotnet/mono/
swift/osascript/tsx/ts-node/bunx/pypy/jruby/raku/luajit/wish/expect/iex).
Deny runs before the operator allowlist, so an alias cannot be allowlisted back
in. Canonicalization only feeds the deny check, so a legit name that ends in a
digit still reaches the normal allowlist check rather than being mis-denied.
Adds validator + integration regressions for versioned/alias runtimes asserting
no DB row and no connection, including the allowlisted-anyway case.
The scheduled-task runner built the agent's tool set from RAG retrieval plus
ASSISTANT_ALWAYS_AVAILABLE. Neither includes bash/python (nor the file tools),
and no keyword hint force-includes them, so a task only saw the shell when the
tool-embedding index happened to surface it. On hosts where that index is empty
or degraded (e.g. a fresh Docker deploy), retrieval returns nothing and the task
agent never receives bash/python — telling the user the shell is disabled even
for an admin owner.
Offer the shell/file group to task agents by default, mirroring the chat agent
where these are on unless a privilege or global setting turns them off. The
existing blocked_tools_for_owner() gate in stream_agent_loop still strips the
whole group for non-admin multi-user owners and only admits it for admins and
single-user (AUTH_ENABLED=false) deployments, so this changes what is offered,
not who is allowed. A crew that defines an explicit enabled_tools allowlist
still has its restriction honored.
Also merge the operator's global disabled_tools setting into the scheduler's
disabled set before composing relevant_tools and before entering the agent
loop, matching what chat already does. Without it, the global tool-disable
contract did not reach unattended scheduled tasks: an admin or AUTH_ENABLED=false
task could still see and call shell/file tools the operator had turned off
globally, since the prompt/schema/execution gates only enforce the disabled
tools passed in.
The non-native (prompted) tool-call path fed tool output back to the model as a plain "[Tool execution results]" user message, bypassing the untrusted_context_message wrapper that THREAT_MODEL.md requires for tool output. That path is what models without native tool-calling (many smaller local models) use, so prompt-injection inside a tool result (fetched page, file read, MCP/email output) could be read as instructions there.
Wrap it via untrusted_context_message("tool execution results", ...), the same hardening already applied to skills (#788) and escalation traces (#275). Also update _recent_context_for_retrieval, which used the old "[Tool execution results]" prefix as a sentinel to keep tool envelopes out of the retrieval query, to recognise the wrapped envelope via metadata.trusted.
The native path keeps returning tool-role messages (a user-role wrapper would break the native tool-call contract); it is covered by UNTRUSTED_CONTEXT_POLICY. Adds tests/test_tool_output_prompt_injection.py.
Fixes#1627.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Added PASSWORD_MIN_LENGTH and RESERVED_USERNAMES to src/constants.py as the
single source of truth. Previously PASSWORD_MIN_LENGTH was hardcoded as 8 in
four route handlers and all three JS validation paths; RESERVED_USERNAMES was
an inline frozenset duplicated in core/auth.py, routes/assistant_routes.py,
routes/research_routes.py, and src/task_scheduler.py.
Added GET /api/auth/policy (unauthenticated) so the frontend reads the real
values from the server instead of hardcoding them in JS.
Added missing empty-username guard to /setup and admin POST /users. Both
returned a misleading 500/409 on whitespace-only input. /signup already had the
check; this makes all three consistent.
docker-compose.yml injects FASTEMBED_CACHE_PATH=${FASTEMBED_CACHE_PATH:-},
which sets the variable to an empty string when the host has not defined it.
FASTEMBED_CACHE_DIR used os.getenv("FASTEMBED_CACHE_PATH", default), and
os.getenv only returns the default when the variable is ABSENT -- so the empty
value won and FASTEMBED_CACHE_DIR became "". os.makedirs("") then raised
[Errno 2] No such file or directory: '', FastEmbed failed to initialise, and
every vector feature (RAG, semantic memory, tool index) silently degraded on
the default Docker stack.
Treat an empty value like an absent one via `os.getenv(...) or default`.
Add a regression test covering the empty, unset, and explicit cases.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix: check-in calendar digest leaks every user's events (no owner scope)
* Seed dtend on calendar events in digest test so the NOT NULL column is satisfied
The outbound UA for web_fetch / web_search was inlined in four places with
two different values and nothing keeping them current: content.py pinned a
mid-2021 Chrome 91 build, and providers.py sent a bare Mozilla/5.0 in three
spots. Some sites serve a degraded or blocked page to a UA that old.
Add WEB_FETCH_USER_AGENT to src/constants.py (env-overridable, matching the
existing Copilot/Kimi UA-constant pattern) and import it in content.py and
providers.py. Default to a current, common desktop UA so pages return their
normal HTML: the market-leading desktop OS (Windows; NT 10.0 covers Windows
10 and 11) and browser (Chrome) on a current stable build. The version is now
bumped in one place.
Service-specific self-identifying agents (Copilot, Kimi, webhooks, cookbook)
are intentionally left separate. Adds a regression pinning the constant shape,
the env override, and a guard against a new inline Mozilla literal in the
search sources.
Closes#4324
When a CardDAV contact matched the search query but had no email
address (only phone numbers), the tool silently dropped it and
returned 'No contacts found'. Fall back to the contact's phone
number(s) so the caller still receives usable information.
Refs: #4178 (the contacts-domain classifier fix that made the model
actually call resolve_contact for contacts queries, surfacing this
pre-existing gap)
* fix(search): add download budgets to web_fetch with truncation notice and hard ceiling
MAX_OUTPUT_CHARS only trims what the agent sees; fetch_webpage_content
buffered and cached the entire response body first, so a large or hostile
URL could pull arbitrarily many bytes into memory and the content cache.
The fetch is now a capped streaming GET (SSRF redirect guard unchanged):
a soft default budget (WEB_FETCH_SOFT_MAX_BYTES, 2 MB), a per-call
override via full/max_bytes on the web_fetch tool, and a hard ceiling
(WEB_FETCH_HARD_MAX_BYTES, 20 MB) that the override can never exceed.
When Content-Length already declares a body over the ceiling the fetch
is refused before any body bytes are buffered. Truncated results carry
truncated/fetched_bytes/total_bytes, the tool output leads with a
partial-content notice telling the model how to re-fetch with full=true,
and the tool schema documents the flag. A truncated PDF is reported as
a budget error since a cut PDF is unparseable. The effective cap is part
of the content-cache key so a truncated fetch is never served to a
full-budget request.
Existing tests that faked httpx.get or the old _get_public_url signature
are adapted to the streaming interface; behavior pins are unchanged.
Fixes#3812
* fix(search): close compressed-body cap bypass and protect the partial notice
Addresses RaresKeY's review on #3955:
- Force Accept-Encoding: identity for the capped fetch. With gzip/deflate the
wire bytes (and Content-Length) can be a fraction of the decoded body, so a
tiny compressed response could pass the hard-cap preflight and then expand
past the ceiling in a single decoded chunk before the streamed cap could
slice it. Identity makes Content-Length the true body size and keeps each
streamed chunk bounded by the network read, so the hard ceiling actually
bounds memory.
- Lead web_fetch output with the partial-content notice and cap the page
title. The notice is the user-facing contract for partial fetches, but the
title is untrusted, uncapped page content; placed ahead of the notice a giant
title could push it past MAX_OUTPUT_CHARS and drop it. The notice now leads
and the title is capped as a second guard.
Adds regressions: the fetch advertises identity encoding, and a truncated
result with an oversized title still surfaces the partial notice.
* fix(search): reject compressed responses that ignore the identity request
Requesting Accept-Encoding: identity is not enough on its own: a server can
ignore it and still return Content-Encoding: gzip, and httpx.iter_bytes would
decode that, so a tiny compressed body could balloon into one decoded chunk
far past the hard cap before the streamed loop slices it (and Content-Length,
the compressed wire length, makes the preflight and size metadata unreliable).
Refuse a non-identity Content-Encoding before reading the body. Adds a
regression where the server ignores the identity request and returns gzip;
the fetch is refused before any body is decoded.
* log(app): add warnings to silent except Exception blocks
- Internal tool auth header failure now logs a warning instead of
silently passing, making auth bypass easier to spot in logs.
- Token last_used_at update failure now logs at DEBUG (fire-and-forget,
non-critical, but useful when debugging token tracking issues).
- Image ownership verification failure now logs a warning so unexpected
access-check errors surface instead of silently allowing the request.
* log(chat_routes): add warnings to silent except Exception blocks
- clear_orphaned_session_endpoint: log before rollback so failures
appear in traces when users see stale/deleted model options.
- _endpoint_has_model (JSON parse): log malformed cached_models instead
of silently treating endpoint as valid.
- _has_any_visible_model (JSON parse): log malformed cached_models
instead of silently returning empty list.
- timezone header parse: log failure so time-zone-related tool bugs
(wrong scheduled times, calendar events) are traceable.
- attachments JSON parse: log failure so silently-dropped attachments
are visible in server logs.
* log(email_routes): add warnings to silent except Exception blocks
- Email alias resolution failure now logs a warning instead of silently
returning an empty list, making broken account configs diagnosable.
* log(document_routes): add warnings to silent except Exception blocks
- Export ZIP request body parse failure now logs a warning so empty
exports caused by malformed requests are diagnosable.
- clear_active_document failure on detach now logs a warning to help
trace doc re-injection bugs like #1160.
* log(agent_loop): add warnings to silent except Exception blocks
- builtin tool overrides load failure now logs a warning so misconfigured
settings don't silently fall back to defaults without a trace.
- Timezone context injection failure now logs a warning to help debug
incorrect scheduled times in agent-created tasks.
- PDF form-backed document detection failure now logs a warning so
broken form-doc UI is traceable to the root cause.
* log(llm_core): add warnings to silent except Exception blocks
- Malformed URL in _is_ollama_native_url now logs a warning so bad
endpoint configs are traceable instead of silently returning False.
- Model list fetch failure now logs a warning with the endpoint URL so
endpoints that silently vanish from the model picker are diagnosable.
* log: pass exception via exc_info instead of string interpolation
* fix(logging): avoid logging raw URLs in llm_core error paths
Drop the raw url/base_chat_url from the Ollama-detection and
model-list-fetch warning logs added by this sweep, since these values
can contain private hostnames, internal IPs, credentials, or other
deployment details.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
* feat(core): abstract runtime path logic for frozen distribution packages
* Address review feedback: revert browser MCP check, persistent data dir default when frozen, and add path tests
_reminder_minutes matched the offset with (?:m|min|minute|minutes)\b and
(?:h|hr|hour|hours)\b. The trailing \b makes the common plural
abbreviations "mins"/"hrs" fail to match (after "min" the "s" is a word
char, so no boundary), so reminder_minutes "5 mins" or "2 hrs" returned
None and the event was created with no reminder, silently.
Widen the two unit regexes and the matching reminder_only description
regex to a strict superset that also accepts mins/hrs. The sibling
duration parser already accepts these forms (it has no \b), so this only
brings the reminder parser in line.
- Agent: pass the open email reader (uid/folder/account/from/subject/body
preview) on every chat submit so 'reply to this' / 'write email saying
hi' route to ui_control open_email_reply with the right UID instead of
inventing a new .md draft. Code-level enforcement (chat_routes strips
create_document + send_email when active_email is set); cross-session
active_doc_id is now trusted instead of being silently dropped.
set_active_email/clear_active_email tool-layer helpers in
tool_implementations.
- ui_control open_email_reply: optional body argument so the agent can
open-and-write in one call; envelope now forwards uid/folder/account/
body/panel through tool_output. Tool description sharpened and the
parser rejects empty bodies on reply/reply-all (forces the agent to
write rather than open an empty draft).
- Email library: search now runs against [Gmail]/All Mail when the
current folder is INBOX (archived emails surface). Whirlpool spinner
+ 'Searching…' placeholder while in flight. Each search result is
stamped with its source folder so clicks open the right email instead
of whatever shares its UID in INBOX. Search no longer re-applies the
same text pill locally (which only checks subject/from/snippet, never
body) so body-only matches don't get dropped after IMAP returns them.
Initial inbox load bumped 100→500.
- Email favorites: 'Favorite (pin to top)' / 'Unfavorite' in both the
card menu and the open-reader more menu, backed by a new
/api/email/flag/{uid}?on=true|false endpoint. Flagged emails always
bubble to the top of the grid regardless of active sort.
- AI reply in doc editor: never overwrites existing draft text or the
quoted history. AI suggestion is prepended; AI-generated 'On …
wrote:' re-quotes are stripped so the original quote isn't visually
edited.
- Cookbook serve: pre-launch GPU driver / has_gpu / install / version-
floor checks (vllm minimax_m2 needs 0.10.0+, deepseek_r1 needs 0.7.0
etc.) before the launch chain starts. Detect 'another model already
running on this host' and offer Stop & launch (with graceful then
force tmux kill helpers, port release wait). Per-vendor deep-link
buttons (vLLM recipe / SGLang cookbook) with hardware hash. Backend
picker is now a custom dropdown with accent-coloured logos for vLLM,
SGLang, llama.cpp, Ollama, Diffusers; same glyphs added next to
package names in Dependencies. Runtime-readiness note moved inside
the panel (green when ready, red when missing) with an × dismiss.
Esc collapses the expanded card; expanded card scrolls when it
overflows; Trust Remote / Auto Tool / Reasoning Parser / Enforce
Eager / Prefix Caching / Expert Parallel / Speculative / MoE Env on
one row (Reasoning Parser auto-detected per model family).
Dtype→Row 1, GPUs→Row 2 (rightmost). Removed redundant GPU 'auto'
input — command builders read from the GPU button strip. Default
cookbook open is Download tab.
- Cookbook hwfit: 'Model (latest)' / 'Model (oldest)' header sorts by
release_date; release dates can be backfilled with the new
scripts/backfill_model_release_dates.py and recipe metadata pulled
with scripts/import_from_vllm_recipes.py against the upstream
vllm-project/recipes catalog (vllm_recipe + min_vllm_version stamped
on entries).
- Calendar: Quick add hint cycles a random Odysseus-themed example per
open (wooden horse Friday, crew muster 10am daily, council on
Ithaca, …). Typing a time like '11pm' in the event title updates
the hero clock live.
- Doc editor: email-mode Reply button (sparkle icon, accent) opens the
same Fast/Full + context popover the email reader uses; Ctrl+Alt+M
toggles markdown preview.
- Memories panel: custom sort picker with per-option icons, default
'Latest', visible Enabled/Disabled toggle text matching the section
description style.
* Agent: make skill-prescribed tools actually callable
The skill index and matched-skill procedures are injected into the
prompt, but tool selection never followed: manage_skills wasn't in the
RAG-selected schema list (so the model substituted manage_memory), and
a matched skill could prescribe tools (grep, read_file) the model had
no schema for. Now:
- manage_skills rides along whenever the owner has any skills indexed
- a Jaccard-matched skill's requires_toolsets join the selection
- viewing a skill mid-turn via manage_skills unlocks its
requires_toolsets for subsequent rounds
- admin-intent turns send _ADMIN_TOOLS schemas, matching the prompt
text _build_base_prompt already advertises
- index_for(active_toolsets=None) no longer hides requires_toolsets
skills from callers that don't know the active set
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Agent: validate skill requires_toolsets against known tools, not TOOL_SECTIONS
grep/glob/ls ship as function schemas without a prompt-prose section,
so gating on TOOL_SECTIONS silently dropped them from a skill's
requires_toolsets.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(research): preserve Discuss spin-off primer during context trimming
trim_for_context() kept only system_msgs[:1] as essential and dropped the
rest under budget pressure. A research "Discuss" spin-off seeds the report
as a system message that sits after the preface system messages, so it
landed in extra_system and was the first thing evicted once the chat grew
— the conversation then lost its grounding and drifted off task.
Treat any system message carrying research_spinoff_from metadata as
essential, alongside the leading system prompt, so the seeded report
survives trimming. maybe_compact already retains all system messages.
Tests: tests/test_context_compactor.py::TestResearchPrimerPreserved
* fix(research): ground Discuss spin-off chats on the seeded report
build_chat_context injected global memory (pinned + hybrid-retrieved) and
personal-doc RAG every turn, keyed off the user-level memory_enabled pref
and a request-scoped use_rag flag — never the session. A research spin-off,
whose primer declares the report the sole knowledge base, thus had
unrelated keyword-matched facts pulled in ("wrong data") competing with the
report; its rag=False flag was also ignored (use_rag defaulted on).
Add _session_is_research_spinoff(sess) (detects the primer research_spinoff_from
metadata; handles ChatMessage and dict forms) and, for such sessions,
disable memory injection and force RAG off.
Tests: tests/test_chat_helpers.py spin-off detection cases
---------
Co-authored-by: Dan (cirim) <claude@cirim.org>
* fix(kimi): resolve Kimi Code API 403 errors and User-Agent restrictions
Kimi Code subscription keys require a whitelisted coding-agent User-Agent to avoid access_terminated_error 403s. This adds User-Agent probing and caching for Kimi Code endpoints.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(kimi): omit temperature for kimi-for-coding API calls
Kimi Code rejects any non-default temperature with HTTP 400, which broke deep research probes and low-temp LLM rounds.
Co-authored-by: Cursor <cursoragent@cursor.com>
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(agent): don't let a materialized default budget defeat context scaling
#1230 scales agent_input_token_budget to the model's context window unless
the user explicitly set a budget, detected via is_setting_overridden(). But
the settings-save path materializes every DEFAULT_SETTINGS key into
settings.json (load_settings merges defaults; handlers persist the merged
dict), so the persisted default 6000 reads as "overridden" and the budget
code takes the min(6000, ctx) branch — silently re-capping long-context
models at 6000 for anyone who has ever saved a setting. This reintroduces
the exact regression #1170/#1230 set out to fix.
Add is_setting_customized() (saved value != default) and gate the scaling
on it instead of mere presence. A persisted default is not a user choice.
is_setting_overridden has exactly one consumer (this budget path), so the
change is contained. Tests cover the materialized-default regression, a
deliberately-chosen budget still being honoured, and the absent-key case.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(agent): rework context-budget fix per review (#4122)
Address RaresKeY's review:
P2 (explicitness): is_setting_customized treated a saved value equal to the
default as "not explicit", which ALSO blocked a user from deliberately pinning
the default budget. Reframe the default value itself as the AUTO sentinel —
agent_input_token_budget == DEFAULT_BUDGET means "scale to the model's context
window", any other value is an explicit cap. A materialized default still reads
as auto (fixing the original regression), and any non-default value the user
chooses is now honoured. Drop the now-unused is_setting_customized helper.
P2 (fallback context): auto-scaling trusted get_context_length() even when it
returned only the bare DEFAULT_CONTEXT fallback (no endpoint-reported / known
window), over-allocating on self-hosted/proxy setups. Add get_context_length_known()
(also returns whether the window was actually discovered); the budget block
passes 0 when unknown so auto-scaling stays conservative instead of inflating to
an unproven window.
hard_max stays auto-only — a deliberate explicit budget wins (#1190); kept that
contract and answered the reviewer's question rather than silently reversing it.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* test(agent): lock the materialized-default budget regression (review on #4121)
Per WGlynn's review on the issue: add an end-to-end regression that saves an
UNRELATED setting (which makes the settings-save path materialize the budget
default into settings.json) and asserts the budget still auto-scales rather than
re-reading as an explicit 6000 cap — locking the exact reopening shut.
To make the test bite the production decision (not just re-derive it), extract
`budget_is_explicit()` into src/context_budget.py and use it from the agent loop.
It keys off value-vs-default (the default is the auto sentinel), NOT settings
presence — which is the whole point, since the save path materializes defaults.
Note: after this PR's rework, is_setting_overridden has ZERO production callers,
so the merged-dict materialization smell can't reach any setting through a
presence check today (WGlynn's durability concern).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(agent): bind the budget context window to its own provenance (review #4122)
RaresKeY caught a correctness bug in the fallback-context guard: stream_agent_loop
kept only the `known` flag from get_context_length_known() and budgeted off the
passed-in `context_length`, which can come from a *different* lookup. Two failures:
- local endpoints are re-queried, so the passed value can be a stale DEFAULT_CONTEXT
fallback while the fresh probe proves the real (smaller) served context — we'd
scale off the stale value;
- callers that don't pass context_length (scheduled tasks, teacher escalation,
skill test runs, bg_monitor) were capped at 6000 even when a long window is
discoverable.
Extract budget_context_for_model() which returns the freshly-probed window when
known else 0, binding the flag to the value it proves; the agent loop uses it.
Regression tests cover the stale-fallback, no-arg-caller, and probe-error paths.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* docs(agent): fix stale budget comments + tighten to the contract (review #4122)
- settings.py: an explicit budget is clamped to the window only — hard_max is
auto-only (#1190); drop the incorrect "and to hard_max".
- is_setting_overridden docstring: drop the stale "adaptive budgets" example;
point value-sensitive callers at context_budget.budget_is_explicit.
- Tighten the budget-block comments to the contract (default = auto sentinel,
non-default = explicit cap, hard_max = auto-only ceiling).
Comment/docstring-only; no behaviour change.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* docs(agent): correct budget issue citations (#1190 → merged #1230/#1273)
The context-budget contract (auto-sentinel, explicit budgets honoured,
hard_max auto-only) merged via #1230 — #1190 was the earlier, closed,
superseded PR. Re-point the contract comments at #1230 (the live source,
already cited for the auto-sentinel two lines up in settings.py).
The configurable hard_max setting (`agent_input_token_hard_max`) was a
reviewer requirement first raised on #1190, omitted from the merged #1230,
and actually added in #1273 — credit #1273 for it and correct the test
comment's history (it previously implied this PR completed the requirement).
Comment/docstring-only; no behaviour change.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
pewdiepie-archdaemon