Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/cirwel/unitaresWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/cirwel/unitares/governance-start)<a href="https://agentmods.dev/commands/cirwel/unitares/governance-start"><img src="https://agentmods.dev/badge/commands/cirwel/unitares/governance-start.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00012 | $0.00856 |
| Opus 5 | $0.00006 | $0.00428 |
| Sonnet 5 | $0.00002 | $0.00171 |
| Haiku 4.5 | $0.00001 | $0.00086 |
Grade A, and why
governance-start scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Under the identity ontology v2 (docs/ontology/identity.md), a fresh process-instance mints fresh governance-identity. Lineage is declared via parent_agent_id, not resumed via token. This command starts a session in that posture.
Start by checking the local workspace cache inventory.
Use the shared helper in this repo:
scripts/client/session_cache.py list
Then run the workspace start guard:
python3 scripts/dev/workspace_closeout.py --start-check
This fails on dirty git state or non-baseline repo-rooted processes from a prior session, and writes a fresh process baseline only when the workspace is clean.
It also classifies workspace isolation: editing the shared/main checkout (or the
deploy worktree) instead of an agent-owned linked worktree is surfaced as an
advisory warning by default. Add --require-worktree to make that a hard
failure (strict mode) once the fleet reliably works in worktrees — the
advisory→strict rollout mirrors the Surface Lease Plane. Rationale:
docs/proposals/worktree-isolation-vs-lease-default.md.
A cached parent_agent_id from a prior session is context, not a lineage
instruction: co-location in this workspace is not lineage, and the prior session
is almost always still a sibling rather than an exited predecessor. Ignore any
legacy continuity_token field for startup; tokens are only for advanced
same-live-owner diagnostic rebinds.
Call onboard() against UNITARES using the strongest honest mode:
- default: a fresh session onboards fresh — pass
force_new=truewith noparent_agent_id - declare lineage only for a real causal event: a dispatched subagent (
parent_agent_id=<dispatcher uuid>,spawn_reason="subagent", usually set automatically by the dispatcher) or a handoff from a finished prior session (parent_agent_id=<prior uuid>,spawn_reason="new_session"). Declaring a currently-live agent as parent is rejected (lineage_coincidental_rejected). - include
model_typewhen the current runtime is clear from context - do not invent a display name unless the user asked for one
start_session(...) is an equivalent alias (same parameters, same rules). Invoking the alias returns the normalized agent-experience envelope — next_action/state_summary first, agent_uuid and client_session_id lifted to the top level, and the full canonical payload (including session_resolution_source and the other cache fields below) under raw_governance.
Do not use bare identity(agent_uuid=<uuid>, resume=true). UUID alone is an unsigned claim and is hijack-shaped under strict identity mode.
Do not use onboard(continuity_token=...) as cross-process resume; after S1-c it returns status=continuity_token_resume_rejected.
After a successful identity() or onboard() response:
- create or update a slotted cache using
scripts/client/session_cache.py set session --slot <client_session_id-or-codex-session-id> --merge --stamp - keep it compact and machine-readable JSON
- include:
schema_version: 2server_urlwhen knownuuidagent_iddisplay_nameclient_session_idsession_resolution_sourceupdated_at
- do not write
continuity_tokenorcontinuity_token_supportedto the cache
When reporting back:
- say whether the identity was freshly created or created with lineage
- if lineage was declared, name the parent UUID prefix
- show the resolved display name or agent id
- note whether continuity is strong or weak
- mention the next useful command:
/checkinfor the turn baseline, then after meaningful milestones/diagnoseif continuity still looks wrong
Do not dump raw JSON unless the user asks for it. Prefer a short interpreted summary.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 74 lines · 12 tokens per session scan A 66ae87ea0e78
governance-start is a command published in the GitHub repository cirwel/unitares (4 stars, last pushed yesterday), licensed Apache-2.0. It adds 12 tokens to every session and 856 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
coder-eval-implement-plan
Implement an approved codereval plan phase by phase with risk-scaled per-phase review, then a final code review.
nyann:diff-profile
Compare two nyann profiles side-by-side and show what changes between them: hooks, branching, CI, documentation, extras, governance. Useful before switching profiles to understand the impact.
review-tasks
A command that reviews a task list for fidelity to its source plans, practical executability, dependencies, and ways to verify completion. It can use project requirements, specifications, designs, plans, and repository files.
fastapi
FastAPI application design and implementation conventions. Use this skill when building, updating, or reviewing FastAPI services, routers, dependencies, request/response schemas, streaming endpoints, or API tests. Trigger on FastAPI-specific work such as path operation design, dependency injection, response models…
atrs
Generate Algorithmic Transparency Recording Standard (ATRS) record for AI/algorithmic tools.
fr-rgpd
You are helping an enterprise architect generate a French CNIL Compliance Assessment — the French-specific GDPR layer applied by the CNIL (Commission Nationale de l'Informatique et des Libertés). Run this after /arckit-eu:eu-rgpd to add French obligations that go beyond the EU GDPR baseline.