Migrated from Method backlog
This issue was created from a legacy filesystem backlog card. GitHub Issues are now the live work tracker; repository docs remain Method evidence.
Source backlog: docs/method/backlog/cool-ideas/PLATFORM_xtask-leash-audit-stub.md
Original lane: cool-ideas
Original legend: PLATFORM
Original backlog card
xtask leash-audit (minimum viable): print the leash state, don't transition yet
Legend: PLATFORM
The argument for building it now
The convention card
docs/method/backlog/cool-ideas/METHOD_leash-files.md deferred the
full xtask leash-audit machinery (trigger/symbol/status
transitions, grep against repo state, automatic graveyard moves)
under the "you only need a Tool when you have two of the thing"
rule.
That rule is correct in steady state. But the 2026-05-30 pre-Phase-2
review surfaced a sharper observation: Phase 2 GREEN is precisely
the moment the first leash should fire. If the audit machinery
does not exist when that moment arrives, the jedit session-port
scaffold becomes barnacle by default.
A minimum version of xtask leash-audit would prevent that
default without committing to the full state-machine yet.
Minimum viable shape
A single subcommand that reads every docs/method/backlog/leash/ *.md frontmatter and prints a structured table.
$ cargo xtask leash-audit
Active leashes (1):
┌────────────────────────────────────────────────────────────┐
│ scaffold: JeditWorldlineSessionPort │
│ repo: jedit │
│ trigger: echo cycle 0025-sessions-as-causal-contexts │
│ phase Phase 2 GREEN │
│ status: active │
│ │
│ symbols still present in jedit (per declared grep): │
│ ✓ JeditWorldlineSessionPort (4 hits in src/) │
│ ✓ JeditWorldlineSessionNotRegistered (3 hits in src/) │
│ ✓ JeditWorldlineId (7 hits) │
│ ✓ JeditTransportSeam (5 hits) │
│ ✓ jeditSessionPort (8 hits in src+spec) │
│ ✓ createInMemoryJeditWorldlineSessionPort (2 hits) │
│ ✓ installed-jedit-eint-bridge (3 hits) │
└────────────────────────────────────────────────────────────┘
What it does NOT do (yet):
- Transition statuses automatically (
active → triggered → deleted). Phase 2 GREEN will fire the first trigger manually;
the state machine grows out of that real example.
- Walk
docs/method/retro/<cycle>/ to detect closed cycles.
Future work after the first transition is recorded by hand.
- Open PRs to move resolved leashes into graveyard. Adds blast
radius; ship the audit first.
- Cross-repo execution. The minimum version runs against one repo
at a time. The leash file declares its target repo; the audit
treats that as a label, not a remote operation.
Why this size
- Reading frontmatter and grepping symbols is ~100 lines of
xtask code; the state machine is the part that grows over time.
- Printing the table is enough to make the leash visible at the
right cadence (CI, weekly review, cycle close).
- A useful tool ships before the trigger fires; the missing
features get added when their absence hurts.
When the full machinery becomes useful
Specifically: when there are at least two active leashes and at
least one of them has crossed its trigger. At that point the
manual "did X close? have the symbols disappeared?" check is paying
real cost and the state-machine investment is justified.
Until then: the minimum version is the leash, the leash is the
minimum version.
Trigger / acceptance
Resolve this card when:
cargo xtask leash-audit reads every
docs/method/backlog/leash/*.md frontmatter and prints the
active leashes with per-symbol grep hit counts. Leashes are
printed in canonical order by scaffold, then repo, and
per-leash symbols are sorted lexicographically.- The grep is run against the leash's declared
repo working
tree (resolved via a configurable repo-root map, or against the
current working directory when no map is provided).
- The command exits non-zero if any leash references a
repo
that cannot be resolved, so misconfiguration is visible.
Companion
docs/method/backlog/cool-ideas/METHOD_leash-files.md — the
convention this audit walks.https://github.com/flyingrobots/jedit/blob/main/docs/method/backlog/leash/jedit-session-port.md
— the external jedit leash that will make this tool immediately
useful.
Migrated from Method backlog
This issue was created from a legacy filesystem backlog card. GitHub Issues are now the live work tracker; repository docs remain Method evidence.
Source backlog:
docs/method/backlog/cool-ideas/PLATFORM_xtask-leash-audit-stub.mdOriginal lane:
cool-ideasOriginal legend:
PLATFORMOriginal backlog card
xtask leash-audit(minimum viable): print the leash state, don't transition yetLegend:
PLATFORMThe argument for building it now
The convention card
docs/method/backlog/cool-ideas/METHOD_leash-files.mddeferred thefull
xtask leash-auditmachinery (trigger/symbol/statustransitions, grep against repo state, automatic graveyard moves)
under the "you only need a Tool when you have two of the thing"
rule.
That rule is correct in steady state. But the 2026-05-30 pre-Phase-2
review surfaced a sharper observation: Phase 2 GREEN is precisely
the moment the first leash should fire. If the audit machinery
does not exist when that moment arrives, the jedit session-port
scaffold becomes barnacle by default.
A minimum version of
xtask leash-auditwould prevent thatdefault without committing to the full state-machine yet.
Minimum viable shape
A single subcommand that reads every
docs/method/backlog/leash/ *.mdfrontmatter and prints a structured table.What it does NOT do (yet):
active → triggered → deleted). Phase 2 GREEN will fire the first trigger manually;the state machine grows out of that real example.
docs/method/retro/<cycle>/to detect closed cycles.Future work after the first transition is recorded by hand.
radius; ship the audit first.
at a time. The leash file declares its target
repo; the audittreats that as a label, not a remote operation.
Why this size
xtask code; the state machine is the part that grows over time.
right cadence (CI, weekly review, cycle close).
features get added when their absence hurts.
When the full machinery becomes useful
Specifically: when there are at least two active leashes and at
least one of them has crossed its trigger. At that point the
manual "did X close? have the symbols disappeared?" check is paying
real cost and the state-machine investment is justified.
Until then: the minimum version is the leash, the leash is the
minimum version.
Trigger / acceptance
Resolve this card when:
cargo xtask leash-auditreads everydocs/method/backlog/leash/*.mdfrontmatter and prints theactive leashes with per-symbol grep hit counts. Leashes are
printed in canonical order by
scaffold, thenrepo, andper-leash symbols are sorted lexicographically.
repoworkingtree (resolved via a configurable repo-root map, or against the
current working directory when no map is provided).
repothat cannot be resolved, so misconfiguration is visible.
Companion
docs/method/backlog/cool-ideas/METHOD_leash-files.md— theconvention this audit walks.
https://github.com/flyingrobots/jedit/blob/main/docs/method/backlog/leash/jedit-session-port.md— the external jedit leash that will make this tool immediately
useful.