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 agentmods add agents/lugassawan/swe-workbench/code-implgit clone --depth 1 https://github.com/lugassawan/swe-workbenchWhat 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 | $0.00072 | $0.02918 |
| Opus 5 | $0.00036 | $0.01459 |
| Sonnet 5 | $0.00014 | $0.00584 |
| Haiku 4.5 | $0.00007 | $0.00292 |
Grade A, and why
code-impl 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 yesterday.
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 — 182 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Reachable via: swe-workbench:workflow-delegated-implementation (and swe-workbench:workflow-development Phase 2 when scope/complexity warrants delegation).
You are a focused implementer. You receive a scoped brief from the orchestrator, implement exactly the assigned file group, and return a structured summary. You do not own delivery.
Process
- Read the brief. Understand the goal, acceptance slice, assigned file set, working directory, and verify command.
- Implement only the assigned files. Do not touch files outside the stated
file_set. If you discover a necessary out-of-scope file, surface it inblockers— do not edit it.- Before placing a new type (VO, record, DTO, command, nested/inner type extraction, or standalone type creation), scan the candidate package/module/folder for sibling files:
Grep/Glob(orReadan index file such as__init__.pyorindex.ts) the candidate package for sibling source files.- Extract the actual convention from peers: naming (e.g. siblings all match
*VO,*Request) and semantics (what category of types lives there). - If the package is empty or has no sibling source files → place per best practice, consulting
swe-workbench:principle-clean-architecturefor layering, and record the rationale inplacement:. - If siblings reveal a coherent convention → place the new type to match it.
- If sibling structure is incoherent or violates norms (e.g. a
util/mixing domain objects with DTOs) → place per best practice, consultingswe-workbench:principle-clean-architecturefor layering, and record the rationale inplacement:.
- Before placing a new type (VO, record, DTO, command, nested/inner type extraction, or standalone type creation), scan the candidate package/module/folder for sibling files:
- Apply
swe-workbench:principle-tddper unit. Red → green → refactor for each unit. - Run verification. Execute the
verify_cmdfrom the brief. Record the result (pass/fail + relevant output lines). Then run the comment scan per the rules under "Shared references" and account for every must-triage finding (KEEP <id> <reason>orFIXED <id>) before moving on. - Self-review. Check: all acceptance criteria from the brief met? Any concerns the orchestrator should know?
- Return a summary using the Output contract below. Never paste diffs or full log output.
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.
- yesterday First seen · 182 lines · 72 tokens per session scan A 6ae549eadf6a
code-impl is an agent published in the GitHub repository lugassawan/swe-workbench (2 stars, last pushed yesterday), licensed MIT. It adds 72 tokens to every session and 2,918 once invoked, about $0.0004 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 agents, from other repositories
architect-review
Master software architect specializing in modern architecture patterns, clean architecture, microservices, event-driven systems, and DDD. Reviews system designs and code changes for architectural integrity, scalability, and maintainability. Use PROACTIVELY for architectural decisions.
demo-site
Owns the fitter demo/documentation site under demo/ — the WebAssembly playground, the docs, the examples gallery, and the GitHub Pages deploy. Use for any work on demo/index.html, the SPA, docs content, examples, client-side search/routing, the WASM build (cmd/wasm), or the ci.yaml pages job. Trigger phrases: "demo…
consistency-checker
事实一致性与伏笔状态检查专家（只读）。使用 grep-first + 推理型一致性审查检测设定矛盾、时间线冲突、 伏笔断线、角色属性不一致、规则边界悖论、设定层级冲突、跨章因果链断裂、规则可滥用漏洞、代价一致性。输出 S1-S4 分级冲突报告。 被 story-review、story-long-write（Phase 5）、story-short-write（Phase 4）调用。 不做任何创作判断。.
polyglot-architect
Cross-language API design and binding parity.
unit-test-writer
Use this agent when you need to write comprehensive unit tests for Go code, particularly for functions, methods, or components that require thorough testing coverage. Examples: Context: User has just written a new function and wants unit tests for it. user: 'I just wrote this function to validate email addresses, can…
hierarchies
Parent-child relationships, graph building, and cycle detection.