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.
npx skills add getmaat/maat --skill maat-retrospectgit clone --depth 1 https://github.com/getmaat/maatWrote 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/skills/getmaat/maat/maat-retrospect)<a href="https://agentmods.dev/skills/getmaat/maat/maat-retrospect"><img src="https://agentmods.dev/badge/skills/getmaat/maat/maat-retrospect/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/getmaat/maat/maat-retrospect"><img src="https://agentmods.dev/badge/skills/getmaat/maat/maat-retrospect.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00032 | $0.00865 |
| Opus 5.5 | $0.00013 | $0.00346 |
| Sonnet 5.5 | $0.00006 | $0.00173 |
| Haiku 4.5 | $0.00003 | $0.00086 |
Grade A, and why
maat-retrospect 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 2mo 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- maat-retrospect — 100% identical, 0 lines differ
How it starts
The opening of the file, as written. The whole thing — 76 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Retrospect: adopt Ma'at in an existing repository
Follow this procedure when maat init has run in an established repository
and the scaffolded documentation does not yet reflect the actual codebase.
You (the agent) supply the judgment; the maat CLI supplies deterministic
facts; the developer supplies the intent. Work in a pair-programming session —
this procedure is interactive by design.
Ground rules
- Never invent intent. Code tells you what; only the developer can confirm why. When context cannot be recovered, write "unknown — inferred from code" rather than a plausible fabrication.
- Never overwrite hand-written content without showing the developer the
change first. Files
maat initskipped are pre-existing and owned by the team. - Decisions derived after the fact are archaeology, not records. Give
them
status: retrospective, and only document decisions that still constrain the codebase today. Do not backfill superseded history. - Small steps: propose, confirm, write, re-check — one document at a time.
Procedure
- Inventory the gaps deterministically.
- Read the
create/skipoutput ofmaat init(re-run it if needed — it is idempotent and never overwrites).skiplines are pre-existing files to reconcile;createlines are scaffolds with placeholder content to fill. - Run
maat checkand collect every finding. This is your work list.
- Read the
- Explore the codebase before asking anything. Entry points, build and test commands, CI configuration, directory layout, recurring conventions, external dependencies. Form hypotheses to verify in the interview, not open questions the code already answers.
- Interview the developer. Cover at least:
- What does this project do, in one paragraph? (project summary)
- What must an agent never do here? What always? (update protocol, conventions)
- Which parts of the architecture are deliberate and constraining, and why? (candidate retrospective ADRs)
- Which existing docs/READMEs are trustworthy, which are stale?
- Reconcile AGENTS.md. If it pre-existed, merge Ma'at's expectations
(project overview, update protocol, links into
docs/) into the existing text — preserve the team's voice and rules. If it was scaffolded, replace the placeholders with interview results. - Fill the scaffolded docs under
docs/(architecture overview, guides, reference). Set honest front-matter:status: draftfor anything unverified,status: currentonly for reviewed content. Wire uprelated_codeso staleness tracking starts working. - Derive retrospective ADRs in
docs/decisions/from the interview — one per living constraint, using the ADR template, withstatus: retrospective. - Close the loop. Run
maat sync, thenmaat check; fix findings until green. Show the developer the final diff. - Ratchet the severities. Where
.maat.yml'schecksection was loosened for adoption, propose tightening (warn→error) now that the corresponding gaps are closed.
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.
- 2mo ago First seen · 76 lines · 32 tokens per session scan A 43aae88a2fb1
maat-retrospect is a skill published in the GitHub repository getmaat/maat (4 stars, last pushed 15d ago), licensed Apache-2.0. It adds 32 tokens to every session and 865 once invoked, about $0.0001 per session on Opus 5.5. 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 skills, from other repositories
sw-do
Work a SpecWeave increment task by task - claim, implement inside its Files, test, commit, done with evidence - alone, unattended or in lanes, then close. Use for "implement", "next task".
done
Close a SpecWeave increment - every task done or skipped, specweave verify green, a fresh-context review, then specweave complete. Use for "close it", "we are done", "finish up".
review
Adversarial review of an increment before it ships, by a fresh context or another model family; every finding cites path:line and is re-verified. Use for "review", "qa", "critique this".
sw-done
Close a SpecWeave increment - every task done or skipped, specweave verify green, a fresh-context review, then specweave complete. Use for "close it", "we are done", "finish up".
glossary
Build the shared language - interview the user term by term and write GLOSSARY.md with what each term means in the business, the data, the code and the UI. Use for "glossary", "ubiquitous language".
app-store-preflight-compliance
Pre-submission compliance scanner workflow for Apple App Store apps. Use when reviewing iOS, macOS, tvOS, watchOS, or visionOS projects (Swift, Objective-C, React Native, Expo) for App Store rejection risks, submission readiness, privacy compliance, or guideline violations.