@ssheleg/agent-stack 0.13.1 → 0.13.3

package/CHANGELOG.md

@@ -1,5 +1,31 @@
 # Changelog
 
+## v0.13.3 — public contract, measurable before it is marketable
+
+The repository now opens with one install path and one concrete request, carries a
+reviewable `SKILL-CARD.md`, portable trigger and behavior evals, and a generated
+1200×630 social preview. CI runs the pinned family skill audit, validates the eval
+schema, watches its planted defect fail, and checks the preview. The eval suite is
+authored, not presented as a model result.
+
+## v0.13.2 — the residue scan stops answering for other runs
+
+`test/residue.py` now tags every workspace with the process group that made it, and the
+gate-wide scan reads only that tag. Scanning the shared `$TMPDIR` by prefix reported two
+things that are not this run's leak: another session's trees — 37,301 entries under the
+shared directory on 2026-08-24 turned a green suite red — and a tree a FAILING case in an
+earlier run kept ON PURPOSE as its evidence, which poisoned every run after it. The split is
+a pure function with fixtures for both directions, and what it excludes is printed rather
+than dropped in silence.
+
+The README told a reader to run commands the published package cannot run: it ships no
+`test/` directory, so `python3 test/validate.py` resolves in a clone and nowhere else. Measured against the
+published tarball on 2026-08-25. Shipping the suite does not fix it — the plants live in
+`.github/workflows/`, which no packaging npm can express puts in a tarball — so the document
+now names where the command runs instead of claiming it, beside a marker the umbrella's
+validator reads. Naming a dead command is this family's own rule; claiming one is the defect.
+
+
 ## v0.13.1 — a citation into the manifesto is a phrase
 
 Five references into the manifesto had rotted, and every rule they named was

package/README.md

@@ -1,5 +1,26 @@
 # agent-stack
 
+[![validate](https://github.com/ssheleg/agent-stack/actions/workflows/validate.yml/badge.svg)](https://github.com/ssheleg/agent-stack/actions/workflows/validate.yml)
+[![npm](https://img.shields.io/npm/v/%40ssheleg%2Fagent-stack)](https://www.npmjs.com/package/@ssheleg/agent-stack)
+[![license](https://img.shields.io/badge/license-MIT-blue)](LICENSE)
+[![site](https://img.shields.io/badge/docs-skills.sshlg.me-8ab0ff)](https://skills.sshlg.me/skills/agent-stack/)
+
+**Build agent loops, harnesses, evals and protocol boundaries that stay inspectable under production failure.**
+
+```bash
+npx skills add ssheleg/agent-stack
+```
+
+Ask: `Design a resumable orchestrator with parallel collection and one convergence checker.`
+
+**[Detailed docs →](https://skills.sshlg.me/skills/agent-stack/)**
+
+**[Docs, and every skill →](https://skills.sshlg.me/)** · [this skill's page](https://skills.sshlg.me/skills/agent-stack/) · [follow @sshlg93 on X](https://x.com/intent/follow?screen_name=sshlg93)
+
+Loads in **DeepSeek Harness** (`dsh`) with **no plugin to write**: it reads the
+Agent Skills standard directly, scanning `~/.agents/skills` — where `npx skills
+add` puts this pack — at rank 500.
+
 Production patterns for building AI agent orchestrators — and for billing the
 LLM access they burn.
 
@@ -191,6 +212,10 @@ teaches you to route around it.
 
 ## Verify
 
+<!-- commands-run-in: a clone -->
+These run **in a clone of this repository**. The published npm package ships no
+`test/` directory, so from an install they are names, not commands.
+
 ```bash
 python3 test/validate.py
 ```

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ssheleg/agent-stack",
-  "version": "0.13.1",
+  "version": "0.13.3",
   "scripts": {
     "test": "python3 test/validate.py && python3 test/plant_guard_test.py"
   },

package/plugins/agent-stack/.claude-plugin/plugin.json

@@ -2,7 +2,7 @@
   "name": "agent-stack",
   "displayName": "Agent Stack",
   "description": "Two skills: agent-orchestrator — tool-calling loops, multi-stage pipelines with checkpoints, provider routing with fallback, four-layer memory, context engineering, plus the wallet side of reselling LLM access; and agent-evals — run/trace/thread evals, judges, and fixtures grown from production.",
-  "version": "0.13.1",
+  "version": "0.13.3",
   "author": {
     "name": "ssheleg",
     "url": "https://x.com/sshlg93"