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 lantisprime/claude-sdlc --skill surgical-editgit clone --depth 1 https://github.com/lantisprime/claude-sdlcWrote 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/lantisprime/claude-sdlc/surgical-edit)<a href="https://agentmods.dev/skills/lantisprime/claude-sdlc/surgical-edit"><img src="https://agentmods.dev/badge/skills/lantisprime/claude-sdlc/surgical-edit.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 | $0.00115 | $0.00755 |
| Opus 5 | $0.00057 | $0.00378 |
| Sonnet 5 | $0.00023 | $0.00151 |
| Haiku 4.5 | $0.00012 | $0.00076 |
Grade A, and why
surgical-edit 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 — 57 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Surgical Edit
Make the smallest possible diff that satisfies the approved plan.
The rules
-
In-scope files only. Before any Edit/Write, read the plan's
In-scope fileslist. If the target file isn't listed, stop and ask: "This file isn't in the plan. Extend scope, or is this the wrong file?" -
In-scope functions only. Before editing within a file, read the plan's
In-scope functionslist. Only modify functions on that list. Adjacent functions — even if they look wrong — must not be changed. -
No ambient cleanup. Do not reformat, rename, reorder imports, or tidy unrelated code. Even whitespace changes in unrelated regions pollute the diff.
-
No speculative refactors. "This would be cleaner if…" belongs in a separate task. Write the idea to
.claude/sdlc/followups/<task-slug>.mdand move on. -
Unrelated bugs → follow-ups. Noticing a bug while working on something else is valuable. Fixing it in this change is not. Log it.
-
Prefer edit over create. New files require justification: a new module, a new test file for a new function, a new template referenced by the plan. Otherwise, extend existing files.
-
Match surrounding style. Indentation, naming, quote style, import order — mirror what's already there. This is not the place to introduce "better" conventions.
The scope-extension protocol
When you legitimately need a file or function not in the plan (you discovered something during implementation):
- Stop editing.
- Propose the extension: "The change requires modifying
auth.py::refresh_tokenbecausevalidate_tokendepends on it. Extend scope?" - Wait for human approval.
- Update the plan's in-scope list with the approved additions and a one-line reason.
- Resume.
This preserves discipline without blocking honest discovery.
Hooks that enforce this
hooks/diff-scope-check.sh— after each Edit/Write, runsgit diff --name-onlyand flags any files not in the plan's in-scope list.hooks/adjacent-function-detector.sh— after each Edit/Write, parses the diff's function-level hunks (viagit difffunction context or tree-sitter perconfig/tools.json) and flags any functions not in the plan's in-scope list.
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 · 57 lines · 115 tokens per session scan A d96252b8027b
surgical-edit is a skill published in the GitHub repository lantisprime/claude-sdlc (3 stars, last pushed 4mo ago), licensed MIT. It adds 115 tokens to every session and 755 once invoked, about $0.0006 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 skills, from other repositories
setup
Guide the user through verifying Grainulator's MCP servers are running, explain what each server does, handle optional dependencies, and initialize a first sprint if none exists. Use when the plugin is first installed, when MCP servers fail to connect, or when the user asks for help setting up Grainulator.
challenge
Adversarial testing of a specific claim. Try to disprove it.
fetch
Size-efficient URL fetch with semantic extraction. Use for ad-hoc web research when you want content without the raw HTML overhead.
harvest
Sprint analytics — type distributions, stale claims, velocity, prediction scoring.
orchard
Multi-sprint orchestration — view dependency graph, dispatch next ready sprint.
blind-spot
Structural gap analysis -- find what the sprint has not considered.