mcpbeat Sign in

Research Ingest Agent Skill

Ingest a CVE writeup, blog post, advisory, or GitHub repo into the wiki - fetch, dedup via sources:, update the right technique/tool page(s), re-index. Generic knowledge only; never client data.

571 tokens
context cost
the whole folder, loaded on every use
1
files
instructions only
0
copies elsewhere
how many repositories repackaged it
320
stars on the repo
on the repository, not the skill itself

Install

one command, takes just this skill from the repository
npx skills add https://github.com/Encod3d-Sec/TORCH --skill research-ingest

What it tells the agent to use

found in the instruction text
WebFetch fetches pages from the network

The instruction itself

3 sections, as written by the author

Research Ingest

Turn an external source (URL or raw/research/ file) into durable wiki knowledge. Synthesise into existing technique/tool pages; do NOT create one page per source. Full workflow: docs/workflows.md (Research ingest / Git repo ingest).

Procedure

  • Fetch the source.
  • URL: defuddle parse <url> --md for clean markdown (install: npm i -g defuddle); fall back to the WebFetch tool.
  • Local: read the file under raw/research/. GitHub repo: clone via WSL only (wsl -d kali-linux -u kali -- git clone <url> /home/kali/<name>), then read source/README.
  • Find the home page. qmd_query "<technique/CVE class>" via wiki-search MCP -> identify the technique/tool page that should hold this. One class = one page.
  • Dedup (skip rule). Read only the target page's frontmatter. If the ingest slug is already in sources:, STOP - already ingested. Otherwise continue.
  • Synthesise the new payload / bypass / CVE chain / detail into the page's existing sections (Methodology, Key payloads, Bypasses and variants). Add concrete commands in code blocks. Reconstruct any image into a code block - never embed images. No em-dashes.
  • Tool page if a new standalone tool appears: create/update wiki/tools/<tool>.md.
  • Record the source. Add the slug to the page's sources: list; bump date_updated. Register URL/path + slug in raw/manifest.md.
  • Re-index + log. python3 scripts/gen_index.py (catalog) and, after a bulk session, qmd update (search index). Append a one-line entry to session/log.md. Run python3 scripts/lint-wiki.py -q - must be clean.

Guardrails

  • Generic only. Wiki holds reusable methodology; client/engagement specifics stay in targets/<eng>/. Reusable default creds -> wiki/cheatsheets/default-credentials.md; reusable request patterns -> api-request-findings.md.
  • Prefer enriching an existing page over creating a new one. New page only for a genuinely new technique/tool class.

Report: page(s) updated + slug added + lint status.

How to use it

Copy the folder

Take encod3d-sec/research-ingest from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.

Install what it needs

The instructions reference npm. Without those the skill loads but fails at the first command.