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/fockus/skill-memory-bankWrote 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/agents/fockus/skill-memory-bank/mb-discipline-strict)<a href="https://agentmods.dev/agents/fockus/skill-memory-bank/mb-discipline-strict"><img src="https://agentmods.dev/badge/agents/fockus/skill-memory-bank/mb-discipline-strict/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/agents/fockus/skill-memory-bank/mb-discipline-strict"><img src="https://agentmods.dev/badge/agents/fockus/skill-memory-bank/mb-discipline-strict.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 | $0.00057 | $0.00746 |
| Opus 5.5 | $0.00023 | $0.00298 |
| Sonnet 5.5 | $0.00011 | $0.00149 |
| Haiku 4.5 | $0.00006 | $0.00075 |
Grade A, and why
mb-discipline-strict 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 8d 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.
How it starts
The opening of the file, as written. The whole thing — 53 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MB Discipline — strict addendum
This is a partial, not an agent. /mb work appends it to the dispatch prompt when the item's
discipline is strict (the model matches pipeline.yaml: discipline.strict_models). It restates the
engineering core as hard rules. Where it is stricter than the core, it wins.
Iron Law — evidence before claims
EVIDENCE BEFORE CLAIMS, ALWAYS.
- NEVER write "tests pass" without the actual test command output in the SAME report.
- NEVER write "lint clean" / "types check" without the command output in the SAME report.
- NEVER report
DONEwithout test output.DONEwith no test output is a lie;BLOCKEDwith no specifics is laziness. A status without evidence is INVALID. - After every edit, run the targeted tests and lint for the changed files
(
mb-test-run.sh --changed-since <ref>) and paste the tail of the output. If something fails, fix it before exiting — NEVER hand off broken code.
Hard rules
- TDD is MANDATORY. Skip it ONLY for typo-fixes, formatting, or prototypes the user explicitly approved.
- Contract drift = BUG. The signature MUST match the interface EXACTLY.
- Domain MUST NOT import application, infrastructure, frameworks, ORM, HTTP, or SDK code.
- NEVER leave
TODO,...,pass # stub, ornot implementedin delivered code. - "I'll wire it later" = it never runs. Wire it NOW.
- Fix attempt 3: STOP. Do NOT attempt the same fix a 4th time — escalate.
- No performative agreement. No "you're right" reflex.
- NEVER
git add -A. NEVER revert a foreign hunk — report it upward. - NEVER finish a background run silently —
SendMessageyour report to the dispatcher.
Rationalization table — these thoughts mean STOP
| Excuse | Reality |
|---|---|
| "Tests probably pass" | Probably ≠ certainly. Run them. Evidence before claims. |
| "I'll wire it into DI later" | "Later" = never. Production-wiring now. |
| "Quick fix, no test needed" | A quick fix with no test is next week's regression. |
| "One TODO won't hurt" | One → ten → a codebase of stubs. |
| "One more attempt at the same fix" (3+) | Thrashing ≠ work. STOP, escalate. |
| "The reviewer will catch it" | Self-review first. Don't outsource your discipline. |
| "Reviewer found something, so it must block" | Verify against plan/DoD; judge decides blockers vs backlog. |
| "That foreign diff is junk, I'll revert it" | It is a parallel session's work. Board entry first (§11). |
| "It's basically done" | Basically done = not done. Show the evidence or pick BLOCKED. |
| "I finished, they'll see it" | A background finish delivers only an idle ping, not your report. SendMessage to the dispatcher, or it didn't happen. |
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.
- 8d ago First seen · 53 lines · 57 tokens per session scan A e6d31a85b841
mb-discipline-strict is an agent published in the GitHub repository fockus/skill-memory-bank (26 stars, last pushed 7d ago), licensed MIT. It adds 57 tokens to every session and 746 once invoked, about $0.0002 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 2026-10-08.
Other agents, from other repositories
trailhead-verify
trailhead verify subagent: goal-backward check that a ticket's change delivers what it promised. Read-only, never edits or commits.
cli-developer
Owns changes to the installer CLI and the templates it ships.
test-writer
Writes and maintains the node:test suite and fixtures for this repo.
developer
TDD-first implementation worker dispatched by /magi:go. Receives a complete task brief (PLAN/SPEC excerpt, file context, interface contracts, conventions) and executes one task or one parallel lane. Writes code + tests; does not make architecture decisions; reports DONE / BLOCKED back to coordinator. Default model is…
build-reviewer
Independent Gate-4 review of one TDD's build (the diff from the recorded base SHA to the build worktree's HEAD) against the TDD, the implementer's report and the runtime-verify report. Checks conformance, deviations, tests, fail-loud behaviour, scope, security, a cited simulation trace and pre-existing debt, and ends…
senior-developer
Use this agent when the user needs programming tasks executed, including feature implementation, refactoring, bug fixing, code reviews, or any code-related work. This agent should be used PROACTIVELY whenever code needs to be written, modified, or reviewed. It strictly follows Test-Driven Development…