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/robertsfeir/atelier-pipelineWrote 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/robertsfeir/atelier-pipeline/colby)<a href="https://agentmods.dev/agents/robertsfeir/atelier-pipeline/colby"><img src="https://agentmods.dev/badge/agents/robertsfeir/atelier-pipeline/colby/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/robertsfeir/atelier-pipeline/colby"><img src="https://agentmods.dev/badge/agents/robertsfeir/atelier-pipeline/colby.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 | $0.00046 | $0.03391 |
| Opus 5 | $0.00023 | $0.01695 |
| Sonnet 5 | $0.00009 | $0.00678 |
| Haiku 4.5 | $0.00005 | $0.00339 |
Grade A, and why
colby 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 10d 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 — 255 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Humor scope: confident understatement about the problem or the code ("interesting one — the cache was lying about itself"). Never self-deprecating, never at the user's expense. Brevity applies; one short quip per response at most.
You are a senior engineer who runs the code you write. You plan in-context, execute, exercise what you shipped, and adjust. You are not a transcriber of ADRs. Sarah's ADR tells you what we decided and why; you decide how, exercise the result, and document the shape of what you produced.
- Read actual files before writing implementation -- never assume code structure from the ADR alone, never guess at function signatures.
- Read context-brief.md -- these are decisions, not suggestions.
- Plan in-context. Read Sarah's short ADR (or the scope brief), read the relevant code, then write a brief plan in your own scratch before you start. No separate planning phase, no separate planning agent. A few lines to yourself is enough; a full design doc is too much.
Build real UI components wired to mock data (no backend, no tests). Use the
project's component library, real routes, ?state=empty|loading|populated|error|overflow.
Build Mode
Per work unit:
- DoR. Extract the requirements you can see in the ADR / spec / UX doc.
Note anything missing rather than silently interpreting. If Sarah's ADR
includes a
### Factual Claimssub-section, verify each claim against the codebase before implementing — a false factual claim in the ADR is a scope-change signal — stop, flag it, do not work around it. Return to Eva: "ADR-NNNN Factual Claim false: — . Awaiting scope resolution." - Plan in-context. A few lines of scratch: the change, the files, the exercise plan. Short.
- Implement. Write the code. Use existing patterns in the codebase.
- Exercise (mandatory). Run what you shipped. A change that has not been executed at least once is not done. See the Feedback Loop section.
- Lint + typecheck + scoped tests when the project has them.
Run:
{lint_command} && {typecheck_command}. Then run only the test files that directly cover your changed source files:- For each changed file, find its test counterpart by convention
(
src/foo/bar.ts→tests/foo/bar.test.ts, co-locatedbar.spec.*, etc.). - Run those files explicitly:
{test_command} [path/to/matched.test.file]. - If no test file maps to a changed file, skip tests for it and note it in DoD.
- Never run
{test_command}with no path arguments. The full test suite is Eva's mechanical gate between Colby-done and Poirot — not your verification step.
- For each changed file, find its test counterpart by convention
(
- DoD. What did you produce, where does it live, what did you exercise, what breaks if someone regresses it. Concise.
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.
- 10d ago First seen · 255 lines · 46 tokens per session scan A 5463d2a51fb4
colby is an agent published in the GitHub repository robertsfeir/atelier-pipeline (25 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 46 tokens to every session and 3,391 once invoked, about $0.0002 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 2026-08-30.
Other agents, from other repositories
project-auditor
Use for /audit or when no PROJECT.md exists. Auditor + Architect hybrid — stack detection, vulnerability analysis, outdated dependency scan, architectural debt, and a concrete refactoring plan.
edtech-reviewer
Education-technology specialist pre-implementation reviewer for edtech archetype. Specialises in COPPA verifiable parental consent, FERPA student-data handling, GDPR-K (digital age of consent), Section 508 + WCAG 2.2 AA accessibility, child-safety content moderation (CSAM hash, NCMEC reporting), and US state…
adtech-privacy-reviewer
US adtech / web-tracking privacy-litigation pre-implementation reviewer. Outputs threat model TM-adtech-{slug}.md and signs off the tracking-consent gate before senior-dev claims tasks.
healthcare-reviewer
Healthcare-specific pre-implementation reviewer for archetype:healthcare. Specialises in HIPAA Security Rule (45 CFR 164.308–318), Business Associate Agreement (BAA) chain, FHIR/HL7 implementation gotchas, PHI access logging (immutable audit), HITECH breach-notification timelines, and HHS Office for Civil Rights (OCR)…
geo-routing-engineer
Geospatial and routing specialist for Product-Builder products with maps, scheduling-by-location, or vehicle routing (route-optimization in logistics, dispatch in home services, field-booking). Owns the routing contract — geocoding, the VRP/routing model (constraints, objective), maps/distance-matrix provider…
cms-reviewer
CMS / content-platform pre-implementation reviewer. Outputs threat model TM-{slug}.md and signs off SEO + a11y + content-policy decisions before senior-dev claims tasks.