AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Self Improving Agent

skill-droodotfoo-agent-skills-self-improving-agent · by DROOdotFOO

>

No reviews yet
0 installs
41 views
0.0% view→install

Install

$ agentstack add skill-droodotfoo-agent-skills-self-improving-agent

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-droodotfoo-agent-skills-self-improving-agent)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
2mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Self Improving Agent? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Self-Improving Agent

Curate session memory, promote recurring patterns to permanent config, and extract reusable skills.

Memory Stack

Three tiers, from most permanent to most ephemeral:

  1. CLAUDE.md (you write) -- Project-wide rules, preferences, conventions. Survives across all sessions.
  2. MEMORY.md (Claude writes) -- Observations, patterns, corrections accumulated during sessions. Review candidates.
  3. Session Memory -- In-context learnings that exist only for the current session. Lost on exit unless promoted.

Commands

| Command | Action | |---------|--------| | /si:review | Scan MEMORY.md for promotion candidates. Show each with recurrence count and recommendation. | | /si:promote | Promote a specific pattern from MEMORY.md to CLAUDE.md or .claude/rules/. Remove from MEMORY.md after promotion. | | /si:extract | Generate a complete skill from a recurring pattern (creates skills//SKILL.md). | | /si:status | Show memory stats: entries in MEMORY.md, staleness, promotion candidates, session observations. | | /si:remember | Capture a specific observation to MEMORY.md with timestamp and context. |

Promotion Lifecycle

See [promotion-lifecycle.md](./promotion-lifecycle.md) for detailed rules.

Summary: discover -> recurs 2-3x -> review flags -> promote to CLAUDE.md or rules -> remove from MEMORY.md.

MEMORY.md Format

## Patterns

- [2026-04-01] Always use `--no-ff` for feature merges (seen 3x)
- [2026-04-03] Tests fail silently when DB not running -- add health check (seen 2x)

## Corrections

- [2026-04-02] User prefers `Result` over panics in Rust (corrected 1x)

## Observations

- [2026-04-05] Build takes 4min -- investigate caching

What You Get

  • A curated MEMORY.md with timestamped patterns, corrections, and observations accumulated across sessions
  • Promotion recommendations identifying recurring patterns ready to graduate to CLAUDE.md or .claude/rules/
  • Extracted skills scaffolded from patterns that recur frequently enough to warrant standalone skill files

Rules

  1. Never promote after a single occurrence -- wait for recurrence
  2. Always show the user what will be promoted and where before writing
  3. Remove promoted entries from MEMORY.md to prevent duplication
  4. Timestamp every MEMORY.md entry
  5. Capture error patterns automatically (see promotion-lifecycle.md for error-capture hook)

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.