the-high-council 0.8.2 → 0.8.3

package/.env.example

@@ -13,11 +13,11 @@ ZAI_API_KEY=
 
 # Default per-run spend ceiling in USD. Overridden per run by --max-usd.
 # A run stops cleanly before any stage that could take it past this, and
-# resumes with a higher ceiling. Unset means $7. Use --max-usd none to run
-# without a ceiling.
+# resumes with a higher ceiling. Unset means $7 for a chain with no paid seat;
+# a paid chain asks you to type its limit (this value is only the suggestion).
+# --max-usd none works for a chain with no paid seat only.
 MAX_USD_PER_RUN=7
 
-# Optional settings (README "Setup" lists them all): COUNCIL_WORKDIR and
-# COUNCIL_MAX_USD_LIMIT for the MCP server, COUNCIL_ALLOW_LOOPBACK_KEY_HOST=1
-# for a local proxy in front of a keyed provider, OLLAMA_API_KEY,
-# COUNCIL_PERSONAS_FILE, AUDIT_HMAC_KEY_FILE.
+# Optional settings (README "Setup" lists them all): COUNCIL_MAX_USD_LIMIT (a ceiling: the MCP server's tools cannot lift it and the limit you type for a paid run, or the total you type for a paid tree (`council tree start --max-usd`, `council tree decide --amend-cap`), cannot go above it), OLLAMA_API_KEY, AUDIT_HMAC_KEY_FILE.
+# Six settings are NOT read from a .env (the CLI prints a notice and ignores them there): set them in the real environment:
+# COUNCIL_ADVICE_SENSITIVITY_FLOOR, COUNCIL_ADVICE_CHAINS_DIR, COUNCIL_WORKDIR, COUNCIL_PERSONAS_FILE, COUNCIL_PII_GATE and COUNCIL_ALLOW_LOOPBACK_KEY_HOST=1 (for a local proxy in front of a keyed provider).

package/CONTRIBUTING.md

@@ -1,68 +1,20 @@
 # Contributing
 
-This project has one maintainer. The bottleneck is reviewing pull requests, not writing code -
-an unscoped PR costs more of that time than it saves. The four shapes below are pre-scoped and
-easy to land; a PR outside them is welcome but will take longer to review, since it needs a scope
-discussion first.
+The most useful help right now: tell us what your runs cost.
 
-## The four smallest landable contributions
+## Send a run cost
 
-1. **A new provider adapter** (`src/providers.js`). Every provider except `mock`/`external` speaks
-   either the Anthropic Messages API or the OpenAI chat-completions shape via `OPENAI_COMPAT`.
-   Adding a new OpenAI-compatible provider is usually one entry in `OPENAI_COMPAT` plus a pricing
-   row in `src/pricing.json`. Test it the way the existing providers are tested: no network calls
-   in `node --test`, real calls are opt-in and cost real money.
+After a paid run, type `council --spend --days 30` in your project folder. Then
+[open an issue](https://github.com/muad-yasin/the-high-council-mcp/issues/new) titled "Run cost: <chain>" with
+that run's line, your version (`council --version`) and roughly how long your task was. Never paste your task,
+your plan or a folder path. Your figure goes into the README's [Pricing](README.md#pricing) table.
 
-2. **A new chain config** (`chains/*.json`). A chain is data, not code - it names which seats run
-   which stages. `council doctor --chain chains/<yours>.json` lints it (the same rules every run
-   applies, including the denied-model and `key-host` rules), and `council doctor` prices it
-   worst-case, both with zero network calls; run both against your new file before opening a PR. Look at `chains/cheap-7-v2.json` or `chains/verify.json` as a starting
-   shape depending on whether you want a debate-and-panel chain or a single-critic chain.
+## Code
 
-3. **A new mock fixture** (`src/providers.js`'s `callMock`, exercised from `test/*.test.js`). If
-   you're adding test coverage for a new code path in the chain itself, the offline `mock`
-   provider needs a new scripted branch (keyed on a `mock-<something>` model name) rather than a
-   real API call. See the existing branches in `callMock` - `mock-unreadable`,
-   `mock-provider-error` - for the shape: a model name that triggers one deterministic scripted
-   reply, documented with which real-world failure it stands in for.
+Bug, docs and typo fixes are always welcome. For anything bigger, open an issue first. Quickest to review: a
+provider adapter, a chain config, a mock fixture or a skill change. `npm test` runs offline for $0 and checks
+the project's rules (no Grok, reasoning never lowered, prompt wording only in `src/roles.js`).
 
-4. **A change to a skill** (`skills/<name>/SKILL.md`). The bar is in [`skills/README.md`](skills/README.md):
-   a good proposal names the failure it prevents, where that failure was seen, and what the rule
-   should say. Skills are generic and public, so a change also passes the same leftover-content
-   check every skill shipped through: no project names, internal paths, private model or session
-   names, or third-party text copied in.
+Two things are settled: bring-your-own-key only, and no claim about output quality until it is measured.
 
-## Before you open a PR
-
-These checks run in `npm test`, so they will fail a PR rather than surprise it in review:
-
-- **No xAI/Grok seat in any chain.** Chain lint's `denied-model` rule refuses it, with no override.
-- **A shipped chain's seats reason at high.** Chain lint's `reasoning-not-high` rule refuses a
-  setting that lowers a seat's reasoning effort.
-- **Prompt wording lives in `src/roles.js` only.** Changing it fails `test/prompt-pin.test.js` on
-  purpose; say in the PR what the prompt now tells the model and why, and the maintainer re-records
-  the pin.
-- **Documents that list what the repo holds are checked against it.** `skills/README.md`,
-  `docs/skills.html` and the landing page (`docs/index.html`) are compared with the repo by tests;
-  a failing one names the document to update.
-- **Run output is never committed.** `runs/` is gitignored; keep it that way.
-
-Two stances, where a PR is closed rather than scoped: bring-your-own-key only (no default key,
-proxy or hosted path), and no claim about output quality until something has been measured.
-
-## Everything else
-
-Bug fixes, docs fixes, and typo fixes are always welcome with no scoping step. Anything larger -
-a new CLI command, a new stage, a change to what a chain's JSON schema can express - should start
-as an issue describing what it's for, before the PR, so review time isn't spent re-deriving scope
-from a diff.
-
-## Running the tests
-
-    npm test                                                  # the whole suite
-    node --test test/chain.test.js                            # one file
-    node --test --test-name-pattern 'parseJson: raw newline'  # one test
-
-Offline, no API key, no network call, no cost. The whole suite is over 2,000 tests and takes
-several minutes; run the files you touched first. `node --test` matches the project's own test
-runner - no additional test framework to install.
+Security issues: [contact@sower-industries.de](mailto:contact@sower-industries.de).

package/PERSONAS.md

@@ -25,7 +25,7 @@ here is a secondary association in a reader's mind, not a legal one - worth havi
 not worth revisiting a decision already made.
 
 A `role.persona` key that is not in the set in use is sent as a bare name, with no voice
-directive. `plan-premium-7` seats two such keys, `hypatia` and `avicenna`, next to the five
+directive. The retired `plan-premium-7` (in `chains/outdated/`) seats two such keys, `hypatia` and `avicenna`, next to the five
 defaults: without a persona file of your own they reach the debate prompt as those names alone.
 
 **You can replace the whole set.** `loadPersonas()` in `src/personas.js` reads an operator-supplied