Update Oracle-to-PostgreSQL Migration Expert: feat: update Oracle-to-Postgres agent model 🤖🤖🤖 - #3364

- Replace open tasks checklist with completed component inventory
- Document deployed workflows with triggers, scopes, and permissions
- Detail custom agents (dev-orchestrator, plan-skeptic)
- Describe state tracking files and their lifecycle
- Add next steps for user testing and initial usage
- Add sync schedule decision (Friday 10:00 AM UTC)
- Comprehensive status update for reference

…nc and contribution guideline monitoring

- Added a new workflow file for the fork-sync-watchdog that schedules a weekly sync from the upstream repository and provides an AI summary of changes.
- Removed the old fork-sync-watchdog.yml workflow, consolidating functionality into the new markdown-based workflow.
- The new workflow checks for changes in contribution guidelines and workflow files, and opens a pull request if there are updates.
- Introduced structured outputs for guideline changes and new workflows, enhancing the maintainability of the fork.

Fork automation: stateless, PR-based tooling for the Oracle-to-PostgreSQL agent

actions/checkout runs with persist-credentials: false even when a
custom token is supplied, so origin has no stored git credentials.
The later 'git push' step to origin then fails with
'could not read Username for https://github.com'.

Add an explicit step that fails fast if FORK_AUTOMATION_PAT is
missing and rewrites origin's remote URL to embed the token via an
env var, plus sets a bot git identity.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

feat: fork automation workflows + fix sync-watchdog PAT push credentials

…e to watchdog summary

- Push step now emits a targeted ::error:: when the PAT lacks the `workflow`
  scope instead of the raw git refusal (first run failed on this).
- Watchdog agent now recommends Keep / Disable / Integrate for every newly
  added upstream workflow file, with disable steps, and flags modified
  upstream workflows for fork impact.
- README: PAT needs public_repo + workflow; troubleshooting row; weekly-sync
  steps. PLAN: constraint, D18/D19, verification log.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

fix(fork-sync): require workflow scope on PAT; add per-workflow triage to watchdog summary

…and recompile lock file

The review suggestion in #5 added '.github/workflows/*.yaml' to the
workflow-detection globs but was applied with the lines de-indented,
which made the `run:` block invalid YAML, and the .lock.yml was never
recompiled. Fix the indentation and regenerate the lock file.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

fix(fork-sync-watchdog): restore indentation of YAML-glob suggestion and recompile lock file

The agent job fetches upstream via git from github.com, which the gh-aw
firewall blocked. Add the 'github' ecosystem to network.allowed and
recompile the lock file.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

Replace the 2-4 sentence prose summary with themed bullets, bold
headlines, section headings, an explicit 'Action needed' line, and
collapsed commit list.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

…-github-domain

fix(fork-sync-watchdog): allow github network egress for agent git fetch

The agent spent the single add_comment quota on a 'test comment' and
then could not post the real summary. Make it explicit that add_comment
is called exactly once with the finished output, and never as a probe.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

…-github-domain

fix(fork-sync-watchdog): forbid test/placeholder add_comment calls

Run logs show the agent completed the analysis and drafted the comment,
then could not post it: gh-aw's own prompt prescribes
'jq -Rs ... | safeoutputs add_comment .' for long bodies, but bash was
allowlisted to 'git *' only, so jq was denied. The fallback
'cat file | safeoutputs add_comment --body -' posted a literal '-' and
exhausted the single add_comment slot.

- Allow 'jq *' in bash tools
- Spell out the exact write-file-then-jq posting command for the
  comment and the issue, and list the transports that must not be used
- Use the schema's 'pr' argument name (pull_request_number is not valid)

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

…-github-domain

fix(fork-sync-watchdog): allow jq so the agent can post the summary body

When the repo owner opens an issue, run the first round of grilling and
planning automatically and hand back a draft PR to work from.

- .github/workflows/fork-issue-planner.md (+ .lock.yml): gh-aw workflow
  triggered by owner-opened issues (fork-automation-labelled issues
  excluded) plus workflow_dispatch for re-runs. It reads the issue and
  the files it touches, classifies scope, and writes a plan containing a
  grilling pass, a concrete implementation plan, a self-adversarial
  skeptic report, open questions and verification commands. Opens a
  draft PR on plan/issue-<N> and comments the open questions on the issue.
- allowed-files restricts the workflow to .github/fork-only/plans/issue-*.md,
  so it can never touch the agent, plugin, skills or any workflow.
- dev-orchestrator now reads the seed plan (and the maintainer's answers)
  in Phase 1 and rewrites it in place; the plan is advisory, not a contract.
- README: architecture diagram, workflow table, development flow,
  PAT scope note, first-run checklist and troubleshooting rows.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

feat(fork-only): plan owner-opened issues into a draft PR

chore(fork-sync): merge upstream main (2026-09-18)

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

docs: record successful fork sync watchdog verification

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

fix: allow planner plan files without review

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

fix: tighten issue planner output contract

Verified via issue #28 -> draft PR #29: correct scope classification,
accurate file reads, full plan sections, one PR + one comment as designed.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

docs(fork-only): mark Fork Issue Planner verification complete

…to .github/agents/

chore: remove completed fork automation plan

Updated model name and corrected description formatting.

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Co-authored-by: PrimedPaul <29710834+PrimedPaul@users.noreply.github.com>

Co-authored-by: PrimedPaul <29710834+PrimedPaul@users.noreply.github.com>

Co-authored-by: PrimedPaul <29710834+PrimedPaul@users.noreply.github.com>

Co-authored-by: PrimedPaul <29710834+PrimedPaul@users.noreply.github.com>

Co-authored-by: PrimedPaul <29710834+PrimedPaul@users.noreply.github.com>

Co-authored-by: PrimedPaul <29710834+PrimedPaul@users.noreply.github.com>