Add DIAGNOSE, CREATE, PERSPECTIVES, FREEFORM intents to Athena's council
classification system. Refine existing AUDIT (add DIAGNOSE boundary),
EVALUATE (improve criteria guidance), and EXPLAIN (remove catch-all status).
Fix bug where prepare_council_prompt defaulted to AUDIT when no intent was
specified — this caused analytical frameworks to be injected into creative
requests (e.g. poetry). Now defaults to no addendum.
Derive validIntents from Object.keys(COUNCIL_INTENT_ADDENDUMS) to prevent
drift between the addendums record and the validation array.
task-output-writer wrote to process.cwd()/.sisyphus/task-outputs/ while
council_finalize read from ctx.directory/.sisyphus/task-outputs/. When
these differ, all output files appear missing.
Fix by passing directory to writeTaskOutput from BackgroundManager, and
adding basePath to council_read. Also removes process.chdir() from tests
that was masking the bug.
Ripgrep's --glob flag silently returns no results when the pattern
contains a directory prefix (e.g. 'apps/backend/**/*.ts'). Models
naturally write patterns this way when exploring unfamiliar codebases.
Extract the static directory segments from the pattern, append them to
the search path, and pass only the glob portion to ripgrep.
- Add council-continuation-enforcer to nudge council members until they
output COUNCIL_MEMBER_RESPONSE, replacing the hacky post-compaction hack
- Add background_wait + background_cancel to council member tool allowlist
so members can properly manage their explore agents before responding
- Update COUNCIL_DELEGATION_ADDENDUM prompt to instruct members to cancel
pending tasks and wait for explore results before final response
- Fix council_finalize path resolution to use project directory
- Remove post-compaction-continuation.ts and recentlyCompactedSessions hack
- Add council-response-checker for detecting response tag in session messages
When models pass relative paths (e.g. 'apps/ios/CleanSlate') to glob/grep
tools, they were passed directly to ripgrep which resolved them against
process.cwd(). In OpenCode Desktop, process.cwd() is '/' causing all
relative path lookups to fail with 'No such file or directory'.
Fix: use path.resolve(ctx.directory, args.path) to resolve relative paths
against the project directory instead of relying on process.cwd().
When models pass relative paths (e.g. 'apps/ios/CleanSlate') to glob/grep
tools, they were passed directly to ripgrep which resolved them against
process.cwd(). In OpenCode Desktop, process.cwd() is '/' causing all
relative path lookups to fail with 'No such file or directory'.
Fix: use path.resolve(ctx.directory, args.path) to resolve relative paths
against the project directory instead of relying on process.cwd().
- Add _isCompleting guard and writeTaskOutput call before status flip
- Propagate writeOutputToFile from LaunchInput to BackgroundTask
- Delete council-result-format.ts and tests (no longer needed)
- 38/38 tests passing
- council_finalize: batch extract, archive, and return council results
- council_read: secure file reader for archived council responses
- background_wait: stripped to metadata-only, returns completed_tasks array
- Removed all council-specific logic from background_wait
- 28/28 tests passing across all changed files
- Remove applyModelThinkingConfig and all callers (OpenCode handles thinking natively)
- Add try/catch error handling to prepare_council_prompt filesystem operations
- Add mode input validation to prepare_council_prompt tool
- Add 2 missing placeholder test cases (RETRY_FAILED_IF_OTHERS_FINISHED, CANCEL_RETRYING_ON_QUORUM)
- Remove redundant 'Do NOT use TodoWrite' prompt section from council member agent
Classify question type BEFORE synthesizing instead of after. Adds 6
signal-based synthesis formats (AUDIT, COMPARISON, DECISION, ROADMAP,
ARCHITECTURE, RESEARCH) so council output shape matches the question
being asked. CONVERSATIONAL path now suggests a natural follow-up.
- Remove COUNCIL_MEMBER_PROMPT from temp file content since it's already
provided as the council member's system prompt via agent config, saving
~350+ tokens per council member per invocation
- Add startup cleanup for stale athena-council-*.md files older than 30
minutes in .sisyphus/tmp/ to handle cases where process exits before
setTimeout cleanup fires
- Remove redundant intermediate permission object spread in createAthenaAgent
- Fix test style: it() -> test(), add #given/#then prefixes per conventions
Both modes now inject explicit instructions: solo warns against subagent usage, delegation provides concrete call_omo_agent examples. Delegation is now the recommended default to reduce context window pressure on council members.
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-opencode)
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
Extract sendPostCompactionContinuation to dedicated file — council members now resume after compaction instead of silently failing. Refresh lastUpdate before async validation in both idle handler and polling path to prevent stale timeout from racing with completion detection.
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-opencode)
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
- Add background_output to council-member allowlist (fixes delegation deadlock)
- Replace empty catch with error logging in prepare-council-prompt
- Remove unnecessary type assertion in agent.ts
- Remove dead hasAgentToolRestrictions function
- Fix incorrect test assertions (undefined vs false semantics)
- Add barrel export for athena module
- Add guard function test coverage (5 tests)
- Add parity test for triple-sync restrictions (9 tests)
- Add session.compacted handler in BackgroundManager to prevent premature
task completion after compaction (defer first post-compaction idle)
- Explicitly block TodoWrite/TodoRead for council members in all sync
points (AgentConfig permission + session tools + prompt instructions)
- Add council member prefix check to todo-continuation-enforcer skip list
to prevent infinite continuation loops on completed council members
- Add optional analysis mode (solo/delegation) question to Athena setup:
solo = thorough but heavier, delegation = fast via explore/librarian
- Allow call_omo_agent in council member allow-list for delegation mode
- Update COUNCIL_MEMBER_PROMPT with TodoWrite prohibition and delegation
addendum for when delegation mode is selected
- Update prepare_council_prompt tool with mode parameter
Council members now use an allow-list (read, grep, glob, lsp_*, ast_grep_search)
instead of a deny-list. Prompt file moved from /tmp/ to .sisyphus/tmp/ so no
external_directory permission is needed. COUNCIL_MEMBER_PROMPT is included in
the temp file for self-contained council member instructions.
Athena saves the analysis prompt to a temp file once, then launches each
council member with a short "Read <path> for your instructions" prompt.
This eliminates repeated prompt text across N task calls while preserving
individual clickable task panes in the TUI.
- Regenerate model-fallback snapshots to include athena agent config
- Fix createBuiltinAgents positional arg index for disableOmoEnv
(shifted from index 12 to 13 by new councilConfig param)
- Fix utils.test.ts, config-handler.test.ts arg positions
- Make directory required in SkillLoadOptions, getAllSkills, and async
skill template resolvers to prevent unsafe process.cwd() fallback
- Remove dead skill export and process.cwd() fallback in skill tool
- Replace kimi-for-coding/k2p5 with kimi-for-coding/kimi-k2.5 in
council-members-generator
- Check context.abort in background-wait and background-output polling loops
- Remove legacy kimi-for-coding/k2p5 from athena fallback chain
- Pass ctx.directory from tool-registry to createSkillTool instead of process.cwd()
- Use case-insensitive casing in duplicate name test to verify actual logic
- Align permission type with SDK AgentConfig pattern (as AgentConfig["permission"])
- Move duplicate-name validation from schema to runtime for graceful fallback
- Place skipped members details before 'end your turn' in council guard prompt
The file no longer contains any fallback/handoff logic after the Athena
NLP removal — only generic terminal-event helpers (isTerminalFinishValue,
isTerminalStepFinishPart). Name now matches content.
The fallback scanned Athena's message text for natural-language handoff
phrases ("switching to Atlas", etc.) and synthetically created a pending
switch when the switch_agent tool wasn't called. In practice this path
never fired in real sessions — Athena always correctly called the tool.
Removes ~135 lines of Athena-coupled code, keeping the generic
switch_agent → apply path fully intact.
Fixes from multi-model council audit (7 members, 19 findings, 9 selected):
- Use parseModelString() for cross-provider Anthropic thinking config (#3)
- Update stale AGENTS.md athena directory listing (#4)
- Replace prompt in appendMissingCouncilPrompt instead of appending (#5)
- Extract duplicated session cleanup logic in agent-switch hook (#6)
- Surface skipped council members when >=2 valid members exist (#9)
- Expand fallback handoff regex with negation guards (#11)
- Remove dead council-member agent from agentSources and tests (#12)
- Make runtime council member duplicate check case-insensitive (#14)
- Fix false-positive schema tests by adding required name field (#18)
ismeth