issues-fs.sgit.ai / admin / comms
Comms: tasks, requests & status
The working channel between the project lead and the site agent, kept in public on the site itself. Updated on every release. Current release: v0.1.1, 22 August 2026. Full history: versions.
Needed from the project lead
| # | Request | Why it blocks | Status |
|---|---|---|---|
| N1 | Name the copyright holder. All nine Issues-FS repositories ship a byte-identical Apache 2.0 LICENSE whose appendix still reads Copyright [yyyy] [name of copyright owner], and none has a NOTICE. Author metadata exists only in pyproject.toml and on PyPI. Apache 2.0 §4(c)/(d) attribution is unanchored across nine repositories and four published packages | Not blocking this site, which publishes its own content CC BY 4.0 and does not mirror upstream documents. Blocking for anybody who wants to reuse the upstream material | waiting on human |
| N2 | Settle the CC BY question for Issues-FS__Docs. One README states that documentation is CC BY 4.0. That sentence is the only occurrence of “CC BY” in the entire corpus, and the repository holding all 59 documents ships Apache 2.0 and says nothing about it. Either add a LICENSE-docs making the dual licence real, or publish the documents under Apache 2.0 and note the difference | Blocking. Until it is settled this site cites paths and quotes rather than republishing upstream documents — which is why The Journey, the concept documents and the two screenplays are catalogued rather than reproduced | waiting on human |
| N3 | Decide what happens to the two empty wheels. issues-fs-service 0.2.0 and issues-fs-service-client-python 0.2.0 each ship an 8-file wheel with no functional code, while their PyPI summaries describe a FastAPI server and a Python client respectively. Fill them, or yank them, or leave them and let the page say so | Not blocking — the site states the position already. It is a statement about a published artefact somebody else owns, so the decision is not the site agent's | open — published as measured, per house convention |
| N4 | Merge Issues-FS__Docs dev → main. Its main is still at Initial commit, 24 commits behind, which is why pip install issues-fs-docs 404s while the README shows the install line and a PyPI badge. The release helper (scripts/gh-release-to-main.sh) is already in the repository | Not blocking this site. It is the single highest-leverage upstream fix available — one command closes a visible 404 | open |
| N5 | Correct the four false README claims at source, so the repositories and this site agree: the storage-backend list, the graph-database integration, the lexicon as a dependency, and the version and test counts. All four, with what was measured | Not blocking. The site publishes the corrections; the READMEs are still the first thing a reader or an agent fetches | open |
| N6 | Mark the superseded lexicon v1. It sits in a folder named already-legacy/ and carries no supersession marker in the file itself, so an agent fetching the path directly has nothing telling it so. One line at the top of the file fixes it | Not blocking. It is the cheapest item on this table | open |
| N7 | Add the Designer role to the role table. It is the eleventh submodule and appears in no README, no role table and no role document. Related: “17 git submodules” is stated twice in CLAUDE.md; there are 19 gitlinks | Not blocking | open |
| N8 | Is the .issues DSL a supported surface, or an internal convenience? Eleven source files, ~55 tests, three live examples and a wired-in repository API say supported. Zero prose documentation and no CLI command say internal. It is currently in the middle: too built to be an experiment, too undocumented to depend on. The page | The site can write the grammar up from the parser and tests either way — but whether to present it as a stable surface is a product call | waiting on human |
| N9 | The reciprocal bridge with graphs.sgit.ai. That site carries an open ask for a bridge page; this site offers the /issues-fs/ half of the trade, and states the boundary — graphs.sgit.ai owns the philosophy, issues-fs.sgit.ai owns the implementation and the origin. Its book review separately asks for an “Issues-FS ontologies & repo” case study; /model/ and /examples/ are this site's answer. Confirm the boundary, so the two sites tell one story rather than two | Not blocking, but it gets more expensive to fix the longer both sites publish | waiting on human |
| N10 | Export the two config JSONs from the flagship graph — node-types.json (12 node types) and link-types.json (10 verb/inverse pairs with domain and range). They are the cheapest credibility available to this site and they also answer an open task on graphs.sgit.ai. The site will not publish a reconstruction of them as though it were the file | Blocks the /examples/ downloads only. The contents are summarised faithfully on the model page in the meantime | waiting on human |
Task board
| # | Task | Owner | Status |
|---|---|---|---|
| T1 | CI pipeline: validate → auto-tag → deploy to Pages, carried over from the sibling sites; validation extended to pull requests; two site-specific gate checks added (every page carries an agent block; no page repeats a corrected claim as fact) | site agent | done v0.1.0 |
| T2 | Theme carried over from graphs.sgit.ai with this site's additions: the package verdict pills, the three-surface cards, the four-operations list, and the file-tree figure | site agent | done v0.1.0 |
| T3 | Shared chrome: nav and footer come from one definition in admin/build/chrome.py, rewritten in place across every page, so the version badge the gate enforces cannot drift. Pages stay hand-written HTML — only the chrome is generated | site agent | done v0.1.0 |
| T4 | /lite/ — the commissioning priority. The specification extracted from §7 of a manual about email in a different repository and given a home: the protocol, a five-minute worked example that did not previously exist anywhere, lite vs full, and the team view and the writer rule. Carries its cross-boundary licence provenance note | site agent | done v0.1.0 |
| T5 | /shipped/ — the navigation job. Eight packages against live PyPI with the wheels unzipped, the dev→main release mechanism that explains every “why isn't it on PyPI”, and the corrections page | site agent | done v0.1.0 |
| T6 | /dsl/ — the first prose documentation of the .issues flat-file DSL anywhere. Describes the shape, the eleven components, the repository API it is wired into, and why it went unnoticed. Deliberately stops short of specifying a grammar that has never been written down | site agent | done v0.1.0 |
| T7 | Write the DSL grammar up from Parser__Issues_File__Line and the ~55 parser tests, and publish the three live .issues files as worked examples with the nodes each produces | site agent | queued — next release |
| T8 | Markdown twins for every page plus llms-full.txt (the sgit.ai convention) — only the front page has a twin today | site agent | queued |
| T9 | Publish the two config JSONs as downloads at /examples/ | site agent | blocked on N10 |
| T10 | Track this site's own build in Issues-FS: run issues-fs init in this repository, commit the .issues/, and render it at /examples/this-site/. The ecosystem already claims the tracker tracks its own development — and it is only partly true today. Costs one command and makes the claim visible | site agent | queued |
| T11 | Reader pages for the upstream documents (raw markdown as source of truth, rendered alongside), as on the sibling sites | site agent | blocked on N2 |
| T12 | Offer graphs.sgit.ai the /issues-fs/ bridge page and agree the boundary wording on both sides | site agent + project lead | queued — after N9 |
How to use this channel
- Human → agent: reply in the working session, or edit this page / open an issue in the repository. Anything added under “Needed from the project lead” with an answer gets actioned next session.
- Agent → human: every release updates this page and versions; requests appear in the N-table above, most-blocking first.
- Decisions log: the site shipped the pipeline before the content, so every release goes through a gate that already works; the corrections are published rather than quietly not-repeated, and the gate enforces them; upstream documents are cited by path rather than mirrored until N2 is settled; the two config files are not reconstructed, because publishing a reconstruction as though it were the file is the exact failure this site was built to correct; and the seven open questions are published unresolved rather than answered by inference.