evidence-pin
Skill0 starsChanged 6 months agoApache-2.0View on GitHub
Citation discipline. Every claim must include a structured evidence pin (artifact + tool + offset/path + raw excerpt). MCP server rejects un-pinned findings — this skill teaches the format.
In this repository13 other files it gives its agents.
- accuracy-reportSkill
- containment-recommenderSkill
- exec-reportSkill
- gap-acknowledgmentSkill
- ir-narrativeSkill
- linux-triageSkill
- macos-triageSkill
- memory-forensicsSkill
- remediation-plannerSkill
- self-correctSkill
- threat-huntingSkill
- triage-orchestratorSkill
- windows-triageSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Every finding MUST be backed by a pin.
- Distinctive: line 84
Do NOT cite a tool you didn't actually call this session
- Very specific: line 85
Do NOT fabricate offsets — the MCP server's verify_excerpt runs server-side and will reject mismatches
- Very specific: line 86
Do NOT pin to /output paths — only /input artifacts are valid sources
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Evidence Pin
Every finding MUST be backed by a pin. The MCP server's finding_record tool rejects records with empty pins[]. Do not try to bypass.
Pin Schema
{
"artifact": "<filename or hive name>",
"tool": "<exact MCP tool name that produced this>",
"locator": {
"type": "registry_path | file_offset | sql_row | log_line | memory_vad | evtx_record_id",
"value": "<the specific locator value>"
},
"raw_excerpt": "<base64 or short hex/text snippet of the underlying bytes>",
"captured_at": "<ISO-8601 timestamp>"
}Examples
Registry
{
"artifact": "NTUSER.DAT",
"tool": "win_registry_get",
"locator": {
"type": "registry_path",
"value": "Software\\Microsoft\\Windows\\CurrentVersion\\Explorer\\UserAssist\\{CEBFF5CD-...}\\Count\\HRZR_PGYFRFFvbaA"
},
"raw_excerpt": "01000000020000005c530000...",
"captured_at": "2026-04-25T22:14:03Z"
}Memory VAD
{
"artifact": "memory.dmp",
"tool": "windows.malfind",
"locator": {
"type": "memory_vad",
"value": "pid=4216 vad_start=0x7ff812340000 vad_end=0x7ff812350000"
},
"raw_excerpt": "4d5a90000300000004000000ffff0000",
"captured_at": "2026-04-25T22:14:09Z"
}EVTX
{
"artifact": "Security.evtx",
"tool": "win_evtx_query",
"locator": {
"type": "evtx_record_id",
"value": "EID=4624, RecordID=18342, Channel=Security"
},
"raw_excerpt": "<EventData>...</EventData>",
"captured_at": "2026-04-25T22:14:15Z"
}Confidence Pairing
Each pin contributes to a finding's confidence. Confidence rules:
| # of Pins | Independent Sources | Confidence |
| 1 | 1 | inferred |
| ≥2 | ≥2 distinct artifacts | confirmed |
| ≥1 | suggestive but not definitive | uncertain |
| 0 | — | INVALID — MCP rejects |
"Independent" means the pin sources don't derive from each other. Two registry keys in the same hive = ONE source. Registry + Prefetch + EVTX = THREE sources.
What NOT To Do
- Do NOT cite a tool you didn't actually call this session
- Do NOT fabricate offsets — the MCP server's
verify_excerptruns server-side and will reject mismatches - Do NOT pin to
/outputpaths — only/inputartifacts are valid sources
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

