diff --git a/CHANGELOG.md b/CHANGELOG.md index e5e47e55a59..c7d88d6d06b 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -11,6 +11,31 @@ from or pin to; the template keeps serving the latest `main` to new forks. ### Added +- **New skill: `video-script`** - turns a repo, product page, or topic into a + recording-ready video script. Receipts-first: every number, address, and date is + verified against the live source in-run, and anything unverifiable is cut or + deferred to a dated "verify before recording" checklist. Standard conventions + throughout (`${var}` selector, `--minutes N` runtime, `soul/` voice, output to + `output/video-scripts/`). `basics` pack, disabled by default; brings the catalog + to **68 skills** (67 -> 68). (#835) +- **New skill: `aeon-update`** - the downstream counterpart to `fork-fleet`. Runs + inside an instance and pulls the parent's shipped framework changes *down* (new + skills, script/harness fixes, workflow and doc updates), landing them as one + reviewable PR - replacing the hand-run rsync-overlay rebase every managed + instance did to stay current with canon. `core` pack, `write` mode, weekly and + disabled by default; catalog **67** (66 -> 67). (#832) +- **New skill: `pack-submit`** - the inverse of `install-skill`. Takes one of the + agent's own local skills and publishes it as a community pack: packages it into a + standalone pack repo, creates and pushes a public GitHub repo to host it, and + opens the registry PR (a README **Community Packs** row plus a matching + `catalog/skill-packs.json` entry in one diff, gated on + `validate-skill-packs.mjs`). `evolution` pack, disabled by default; catalog + **66** (65 -> 66). (#831) +- **Community skill pack registered: Skim Clean Reads** (`skim-read`) - reads any + URL as clean markdown via Skim's x402 endpoint, roughly 4x smaller than the raw + HTML so downstream model steps burn far fewer tokens ($0.002 USDC on Base per + read, no API key). Listed in the Community Packs table and + `bin/install-skill-pack --list`. (#829) - **New notification channel: Buzz.** `./notify` can now post to a [Buzz](https://buzz.xyz) channel (Block's self-hostable Nostr-relay workspace) alongside Telegram / Discord / Slack / email. Unlike the bearer-URL webhooks, @@ -98,6 +123,16 @@ from or pin to; the template keeps serving the latest `main` to new forks. ### Changed +- **README brand refresh.** The `.github/README.md` was rebuilt around an animated + hero and section-banner art with a pill navigation, trimmed to MiroShark prose + density, and the longer-form detail was relocated into `docs/` (Configuration, + Skill packs, Showcase). Adds a "compare vs Claude Code, Hermes, and OpenClaw" + pointer and a Community section. Presentation only - no capability change. (#833, + #834, #836-#838, #840-#842, #844-#857) +- **`memory-flush` hardened.** The memory GC pass now stamps the consolidation date + on every flush (so a live, repeatedly-flushed store no longer reads as an + untouched template), scans an adaptive window instead of a fixed three days, + de-duplicates before writing, and rotates its logs. Body-only skill edit. (#828) - **Adopted the Agent Skills open standard; removed OKF globally.** All 65 skills now use spec-form frontmatter (only `name` + `description` top-level, everything else nested under `metadata:`, block-style lists) and pass the official @@ -129,6 +164,9 @@ from or pin to; the template keeps serving the latest `main` to new forks. ### Fixed +- **Stale "Proof of work" numbers corrected** on the README to match + aeon.fun/security and `ECOSYSTEM.md`: ~2M stars secured, 69 repos, 68 ecosystem + products (community packs stay 12). (#843) - **Read-only persistence contract corrected.** Docs told read-only skills they could persist to `memory/` during a run, but the OS sandbox write-locks the whole workspace on all six harnesses, so those writes silently fail. `CLAUDE.md` and @@ -394,6 +432,9 @@ from or pin to; the template keeps serving the latest `main` to new forks. ### Maintenance +- Webhook Dependabot batch (undici, `@opentelemetry/core`) plus two CI changes: + dropped the daily cron from Setup Telegram Commands, and stated the egress-parser + scope with a lockfile-coverage gate. (#826, #827, #839) - 2 dependency bumps (wrangler in the webhook; the dashboard group) plus a new CI eyebrow capability-integrity gate for skills. (#809, #810, #815) - CI green-up after the Telegram owner-gate, plus a dashboard PacksPanel diff --git a/docs/aeon-setup.md b/docs/aeon-setup.md index 372b4302a81..867d0e834c1 100644 --- a/docs/aeon-setup.md +++ b/docs/aeon-setup.md @@ -9,7 +9,7 @@ Aeon ships an **[Agent Skill](https://code.claude.com/docs/en/skills)** (`SKILL. It's a portable skill, not tied to one product: [Claude Code](https://claude.com/claude-code) is the primary host (it auto-loads skills from `.claude/skills/`, so it's zero-install there), but the skill is just a `SKILL.md` in the open Agent Skills format — it works with **Claude Code, Codex, Hermes, or OpenClaw**, and any other agent tool that supports skills. -> **This is not an Aeon skill.** The 65 skills under [`skills/`](../skills) run unattended on GitHub Actions. **This** one runs inside *your* agent session on *your* machine — it's the operator's assistant for configuring the instance, not something that runs on a schedule. It doesn't appear in `aeon.yml`, the packs, or the catalog. +> **This is not an Aeon skill.** The 68 skills under [`skills/`](../skills) run unattended on GitHub Actions. **This** one runs inside *your* agent session on *your* machine — it's the operator's assistant for configuring the instance, not something that runs on a schedule. It doesn't appear in `aeon.yml`, the packs, or the catalog. ## What it does diff --git a/docs/assets/hero-animated.svg b/docs/assets/hero-animated.svg index 5698f4ac678..af5f5813dc7 100644 --- a/docs/assets/hero-animated.svg +++ b/docs/assets/hero-animated.svg @@ -71,7 +71,7 @@ - 67 skills + 68 skills diff --git a/docs/examples/README.md b/docs/examples/README.md index b6c067f4283..c68f00774cb 100644 --- a/docs/examples/README.md +++ b/docs/examples/README.md @@ -1,6 +1,6 @@ # Aeon integration examples -Aeon ships an [MCP server](../apps/mcp-server/) so any Claude client can call its 65 skills. The scripts here are the shortest possible "first call works" demos — build the server, run `python `, get a real Aeon output back. +Aeon ships an [MCP server](../apps/mcp-server/) so any Claude client can call its 68 skills. The scripts here are the shortest possible "first call works" demos — build the server, run `python `, get a real Aeon output back. | File | Stack | Skill called | What it shows | |------|-------|--------------|---------------| diff --git a/docs/skill-packs.md b/docs/skill-packs.md index e02ad8416c8..6ec720e41aa 100644 --- a/docs/skill-packs.md +++ b/docs/skill-packs.md @@ -1,6 +1,6 @@ # Skill packs -Aeon ships **65 skills**, but most forks only ever run a handful. Packs make +Aeon ships **68 skills**, but most forks only ever run a handful. Packs make that manageable: by default the dashboard shows **Core** (what makes Aeon different) and **Basics** (simple skills you can run right now) — everything else is grouped into **packs** that stay hidden until you enable them. @@ -71,8 +71,8 @@ Pack key = category. Six packs, no empties; three are shown by default. | Pack (`category`) | What's in it | count | |---|---|---| -| **Core** (`core`) | Fleet coordination, self-configuration, liveness, memory + reporting. Shown by default; not removable. | 11 | -| **Evolution** (`evolution`) | The self-improvement loop — authors, evolves, installs, and heals its own skills. Shown by default. | 8 | +| **Core** (`core`) | Fleet coordination, self-configuration, liveness, memory + reporting. Shown by default; not removable. | 12 | +| **Evolution** (`evolution`) | The self-improvement loop — authors, evolves, installs, and heals its own skills. Shown by default. | 9 | | **Basics** (`basics`) | Simple, immediately-runnable skills — one approachable entry per area. Shown by default. | 17 | | **Dev & Code** (`dev`) | PR/issue triage, review, merges, changelogs, repo monitoring, security scanning, app deploys. | 10 | | **Crypto & Markets** (`crypto`) | Token/DeFi/prediction-market monitoring, narrative tracking, on-chain forensics + automation, Uniswap v4 hook deploys. | 14 |