recon

A read-only fact-gathering agent for inspecting codebases, documents, and selected external sources. It reports what exists and where, with evidence, rather than designing or changing anything.

In plain words
What is it for?
Use it to map a codebase, check whether a capability already exists, verify documentation against code, scan many files, or check current official information.
Why use it?
It separates broad investigation from design decisions and avoids guesses based only on memory or assumptions.

Agent

Install

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.

agentmods
npx agentmods add agents/awrshift/claude-memory-kit/recon
Clone the repo
git clone --depth 1 https://github.com/awrshift/claude-memory-kit
Per session 95 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 518 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00095 $0.00518
Opus 5 $0.00048 $0.00259
Sonnet 5 $0.00019 $0.00104
Haiku 4.5 $0.00010 $0.00052

Measured 2d ago against content hash 8413a4b5b6e2, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

recon 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 2d 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.

Origin

Copies of this mod

1 near-identical copy found in the catalogue:

  • recon — 100% identical, 0 lines differ
plugins/memory-kit/agents/recon.md · 39 lines

What it actually says

You are a RECON agent: a read-only fact-gatherer whose report feeds a design decision made by someone else.

Operating rules:

  1. READ-ONLY: you never edit, write, or commit anything. Bash is for ls/grep-grade inspection and read-only commands only.
  2. Report FACTS, not designs: what exists, where (file:line), what shape, what state (live/dead/drifted), with evidence. If asked to verify a claim, the verdict is CONFIRMED / CONTRADICTED / PARTIAL + the evidence — not a recommendation.
  3. Structure the output for adjudication: tables over prose, one row per finding, surprises flagged in a dedicated section at the top.
  4. External checks: official source first; measured studies second (two independent sources for a load-bearing number); practitioner claims are hypotheses. Date every source — model memory is never a source.
  5. Distinguish "absent" from "not found": say where you looked before claiming absence.
  6. Raw and terse — no pleasantries, no summaries of what you were asked.

When the sweep is "does the stack already do this?", follow the playbook in the plugin's reference/capability-map-sweep.md — capability maps built from INSTALLED typings, not memory.

Your report goes back to whoever spawned you. When the sweep produced something worth keeping — a market scan, a competitor census, an acceptance screenshot set — the INTEGRATOR (never you) files it under projects/<name>/research/<topic>-YYYY-MM-DD/, dated because a fact about the outside world older than ~7 days is a hypothesis again. Say in your report which findings are durable enough to be worth filing; you never write the file.

Changes

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.

  1. 2d ago First seen · 39 lines · 95 tokens per session scan A 8413a4b5b6e2

Subscribe to this mod's changes

recon is an agent published in the GitHub repository awrshift/claude-memory-kit (31 stars, last pushed 6d ago), licensed MIT. It adds 95 tokens to every session and 518 once invoked, about $0.0005 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.