Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsBlogPro
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges
  • Chrome Extension
  • Skill Manager

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Learnings

ASecurity

The project's episodic memory: a timestamped ledger of rules paid for with real outages and near-misses, one entry per incident. Load WHENEVER you write or review a DB migration or schema change, touch deploy/release workflows (.github/workflows/deploy-*, promote.yml, vercel config), plan a promote/release, respond to a prod incident, or when another skill references a learning. ALSO load after resolving any incident or near-miss: it must leave a new ledger entry in the same session.

20,239 stars
0 votes
0 copies
0 views
Added 9/29/2026
ai-agentsgobashgit

Security Analysis

A100/100

Pro scans all 20 files and shows the line behind each finding

Scanned 9/29/2026

$npx -y skills add kortix-ai/suna --skill learnings --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Learnings?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Learnings
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/kortix-ai-learnings/badge)](https://www.skillsdirectory.com/skills/kortix-ai-learnings)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
Files
SKILL.md
---
name: learnings
description: "The project's episodic memory: a timestamped ledger of rules paid for with real outages and near-misses, one entry per incident. Load WHENEVER you write or review a DB migration or schema change, touch deploy/release workflows (.github/workflows/deploy-*, promote.yml, vercel config), plan a promote/release, respond to a prod incident, or when another skill references a learning. ALSO load after resolving any incident or near-miss: it must leave a new ledger entry in the same session."
---

# Project learnings: the episodic ledger

Every rule this repository paid for with downtime or a near-miss is one entry in a
timestamped, append-only ledger. The ledger is the full history, in the order it was
learned.

| Path | What it is |
| --- | --- |
| `MEMORY.md` | The index. One line per entry, newest first: recorded time (UTC) and the rule. Generated. |
| `entries/<YYYY-MM-DDTHHMMSSZ>-<slug>.md` | One entry. The filename timestamp is when it was recorded. |
| `scripts/new-entry.sh` | Starts a new entry stamped with the current UTC time and rebuilds the index. |
| `scripts/index.sh [--check]` | Rebuilds `MEMORY.md` from `entries/`, or fails if it is stale. |

Entries recorded before 2026-09-25 were split out of the old single-file register. Their
`recorded` time is the commit that first wrote them (`commit` in the frontmatter).

## Recall: before you touch a surface

1. Search the index and the full text for the surface you are about to change:
   ```bash
   grep -i '<keyword>' .agents/skills/learnings/MEMORY.md
   grep -ril '<keyword>' .agents/skills/learnings/entries/
   ```
   Use the words of the surface: `migration`, `CONCURRENTLY`, `deploy-dev`, `promote`,
   `wake`, `Stop`, `gateway`, `preview`, the file or route name.
2. Read every matching entry in full. Obey its **Rule**, and keep its **Enforcement** green.

Done when every entry that names the surface you touch has been read.

## Record: after an incident or near-miss resolves

Record in the same session. An incident that leaves no entry is not finished.

1. Start the entry:
   ```bash
   .agents/skills/learnings/scripts/new-entry.sh "<the rule, as an imperative>" [incident-date]
   ```
2. Fill the four sections the script writes:
   - **Rule:** the imperative a developer can obey while coding.
   - **Trigger surface:** what someone is doing when it applies.
   - **Incident:** date, version or PR, blast radius, in one to three lines.
   - **Enforcement:** the test, lint, or CI gate that goes red when the rule breaks. An
     entry with an enforcer is a fact. One without is a TODO: name the enforcer to build.
3. Commit the entry and `MEMORY.md` in the same commit.

Done when `scripts/index.sh --check` exits 0 and the entry is committed.

## Ledger rules

- **Append-only.** After an entry merges, its text does not change. When a rule changes,
  or an Enforcer moves or is retired, record a new entry. Put
  `supersedes: <entry filename>` in its frontmatter and state what changed.
- **One entry, one file, one timestamp.** `recorded` in the frontmatter matches the
  filename. It is the time of recording, never the incident date. The incident date goes
  in `incident_date`.
- **`MEMORY.md` is generated.** Run `scripts/index.sh` after any entry change. It merges
  with git's `union` driver, so two branches that each add an entry do not conflict. A
  GitHub squash merge keeps a branch's index line where the branch put it, so the order
  on `main` can drift until the next `index.sh` run. `index.sh --check` and
  `tests/unit/learnings-ledger.test.ts` compare the lines, not their order: they fail on a
  missing, extra, or retitled index line, or a malformed entry.
- **Keep an entry short: about 8 to 20 lines.** Deep detail belongs in the PR body.
  Cite the PR number (`PR #1234`).
- **An entry is self-contained.** The ledger in `entries/` is the only incident record.
  Do not write a separate incident, RCA, spec, or runbook file, and do not link one.
  The repository has no `docs/` tree, and the pre-commit hook rejects new files under it.
- **Synthetic identifiers only.** An entry is committed text, so the AGENTS.md customer-data
  rule applies. Write "a customer", "an enterprise workspace", `<session_id>`.

Attribution

kortix-aikortix-ai
View sourceSee grades on GitHubMore from kortix-ai →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like caveman", "be brief", "less tokens". Stays on until "stop caveman" or "normal mode".

1100021 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

698621 votes

Writing Skills

Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dep...

3931 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3421 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Amp, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Grok Build, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

741 votes
View all in ai-agents →