Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
49 commits
Select commit Hold shift + click to select a range
a82622b
chore: carry three issue captures off main onto a branch
REPPL Sep 9, 2026
a8e1695
chore: record the sub-agent transcript capture gap and its design
REPPL Sep 9, 2026
13eafec
feat: the transcript record carries sub-agent lineage in explicit fields
REPPL Sep 9, 2026
4325d81
feat: sub-agent transcripts are staged when the agent stops
REPPL Sep 9, 2026
0daac9a
chore: record four findings from the sub-agent capture review
REPPL Sep 9, 2026
319da67
feat: history migrate repairs the composite records, history ingest t…
REPPL Sep 9, 2026
9af9a30
feat: history reconstruct emits one agent-readable artefact and its t…
REPPL Sep 9, 2026
d751109
fix: transcript text cannot forge the reconstruction artefact's struc…
REPPL Sep 9, 2026
8d5eadb
fix: staged transcripts stop living indefinitely, and say so when the…
REPPL Sep 9, 2026
a15d214
chore: ship sub-agent capture, recovery and reconstruction as three i…
REPPL Sep 9, 2026
62fb083
fix: repoint the closed specs' links at the intents they now name
REPPL Sep 9, 2026
636731a
chore: rescue three records that existed only in a worktree
REPPL Sep 9, 2026
fb90fc8
fix: nothing quoted from a session can begin a line of its own artefact
REPPL Sep 9, 2026
a07f05e
chore: record 29 findings from an autonomous-run field experiment
REPPL Sep 10, 2026
db894a7
chore: sharpen the merge-conflict finding on its reporting session's …
REPPL Sep 10, 2026
0f9749b
chore: record a third session's autonomous-run findings, one of them …
REPPL Sep 10, 2026
f05ebbd
chore: refine the audit-attestation finding, including this repo's ow…
REPPL Sep 10, 2026
0d7e4f1
chore: sharpen the refused-capture finding with its reporting session…
REPPL Sep 10, 2026
434ce37
chore: close the field experiment with its last two findings
REPPL Sep 10, 2026
7856e63
Merge v0.8.0 into the sub-agent transcript capture branch
REPPL Sep 12, 2026
f854118
fix: clear the release blockers the field experiment surfaced
REPPL Sep 12, 2026
ac474ef
chore: resolve the three store and surface findings the previous comm…
REPPL Sep 12, 2026
14a76a7
chore: defer fifteen findings out loud, each with its reason
REPPL Sep 12, 2026
56f5cd7
fix: a dangling install link and an absent plugin manifest stop being…
REPPL Sep 12, 2026
5c8300b
fix: a declaration vouches for a binary only when nobody else can rew…
REPPL Sep 12, 2026
995fd6c
chore: recalibrate three reading windows against the corpus this cycl…
REPPL Sep 12, 2026
fa77546
chore: four deferred capabilities become intent drafts
REPPL Sep 15, 2026
19348b5
fix: identity redaction masks the spans the detector flagged, not eve…
REPPL Sep 15, 2026
d4c790f
chore: three of the deferred findings are ruled and become records
REPPL Sep 15, 2026
9dd0ed6
feat: a record id written in prose must name a record that exists
REPPL Sep 15, 2026
bfbafa0
feat: a commit that names an issue declares its relation, and a listi…
REPPL Sep 15, 2026
47d733a
feat: an intent owns one or more specs, and it ships when its last sp…
REPPL Sep 15, 2026
9bbcc9a
fix: bind the plugin cache to a home-scoped attestation before promot…
REPPL Sep 15, 2026
0747b32
chore: resolve the owned-copy attestation finding and record the ruling
REPPL Sep 15, 2026
3ed206a
fix: the cache promotion hashes the artefact against the attested val…
REPPL Sep 15, 2026
bc4193b
fix: the bootstrap writes the attestation through a mktemp file, neve…
REPPL Sep 15, 2026
3dfeb84
fix: the attestation's data_dir carries no control character, on the …
REPPL Sep 15, 2026
4036187
Merge remote-tracking branch 'origin/main' into feat/sub-agent-transc…
REPPL Sep 15, 2026
fc5c7e2
chore: recalibrate two reading windows, retire spent baseline entries…
REPPL Sep 15, 2026
ae8846b
Merge remote-tracking branch 'origin/main' into feat/sub-agent-transc…
REPPL Sep 15, 2026
e52d660
Merge branch 'fix/owned-copy-attestation' into the release branch
REPPL Sep 15, 2026
4c8d700
chore: answer the eight findings the merges from main brought into th…
REPPL Sep 15, 2026
241beff
fix: the cache attestation is read only from a record and a home the …
REPPL Sep 15, 2026
19da8f6
chore: record the outbound text check the cycle shipped
REPPL Sep 15, 2026
8f5933f
docs: the sample board shows its presence line, and the mode page say…
REPPL Sep 15, 2026
dca2d77
chore: cut v0.9.0
REPPL Sep 15, 2026
8c4fc71
chore: record the v0.9.0 semantic gate receipts for dca2d779
REPPL Sep 15, 2026
3a41a95
chore: the identity-span test's fake tokens are fixtures the history …
REPPL Sep 16, 2026
e84a40d
docs: intent-auditor 0.3.1 announces the provenance clarification it …
REPPL Sep 16, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
177 changes: 142 additions & 35 deletions .abcd/config/reading-presets.json
Original file line number Diff line number Diff line change
Expand Up @@ -5,11 +5,35 @@
"comment": "The object set is Iteration 1's shipped state (framework 13): the fifteen workstream intents, their fifteen specs, and the packages and pages they delivered that the deny list does not exclude. The kinds are every kind the readings companion's section 5.2 lists for this position - brief current text, the glossary, the disciplines, the specs, and the shipped tree where one exists - with source and test items marked unscanned in the manifest; no intent kind stands here, because neither design document lists the intents in the widening object and itd-194 withdraws the shipped row from the position. The window is the smallest ten-thousand boundary that leaves at least one per cent headroom above what this entry measured on the commit named. The rule is stated because the previous one was not a margin at all: the next boundary above the measurement, whatever room that happened to leave, put detection 63 tokens under its declaration and the entry was breached twice in one day. A boundary is a round number; the headroom has to be asked for separately. The declaration is over the 200,000-token target by a factor of four and inside a reader with a million-token window; the size report states the measured figure on every assembly at this position. A kind that proves useless leaves this entry by a commit that records why: measured here, without test it is about 530,000 estimated tokens and without source and test about 297,000, both still over the target. Changing any part of this entry is a commit, reviewed and inside the dirty gate.",
"object": {
"records": [
"itd-177", "itd-178", "itd-179", "itd-180", "itd-181", "itd-182",
"itd-183", "itd-184", "itd-185", "itd-186", "itd-187", "itd-188",
"itd-189", "itd-198", "itd-199",
"spc-55", "spc-56", "spc-57", "spc-58", "spc-59", "spc-60", "spc-61",
"spc-62", "spc-63", "spc-64", "spc-65", "spc-66", "spc-67", "spc-68",
"itd-177",
"itd-178",
"itd-179",
"itd-180",
"itd-181",
"itd-182",
"itd-183",
"itd-184",
"itd-185",
"itd-186",
"itd-187",
"itd-188",
"itd-189",
"itd-198",
"itd-199",
"spc-55",
"spc-56",
"spc-57",
"spc-58",
"spc-59",
"spc-60",
"spc-61",
"spc-62",
"spc-63",
"spc-64",
"spc-65",
"spc-66",
"spc-67",
"spc-68",
"spc-69"
],
"paths": [
Expand All @@ -25,27 +49,66 @@
"commands/intent.md"
]
},
"kinds": ["brief-section", "glossary-term", "discipline", "spec", "doc", "config", "source", "test"],
"kinds": [
"brief-section",
"glossary-term",
"discipline",
"spec",
"doc",
"config",
"source",
"test"
],
"window": {
"tokens_est": 850000,
"measured_tokens_est": 833115,
"measured_bytes": 3207493,
"measured_at": "cd5e6e19489ab3be4828296b02f945fd71eab4dc"
"tokens_est": 900000,
"measured_tokens_est": 890341,
"measured_bytes": 3427815,
"measured_at": "40361877c4006c7eca3e6fd0cfd56a66d0f59fc8"
}
},
"entailment": {
"comment": "The object set of the other three entries, extended by the ten Iteration 2 intents the record adds to it. The kinds are the claim record read against the constraint sources, which is what the readings companion's section 6.2 names as this position's object: it holds no tree, so no doc, config, source or test kind stands here and the object set's delivered paths hand this position nothing. The ten Iteration 2 intents are NAMED here rather than reached by admission: the maintainer ruled on 2026-09-02, at the Phase A review, that this position is handed the object set only, so the drafts and planned rows narrow by this record list exactly as the shipped row does and a draft or planned intent travels only when this entry names it. The companion's admissibility is a permission, not the scope, and it survives as the `admit_drafts_and_planned` switch, which this entry does not declare and which defaults to off; with it on, the position would be handed every draft and planned intent in the repository, which on the tree this was measured over is 147 projected intents. Several of the ten are shipped by the time Phase A completes and travel through the shipped row, which is the same object set either way. The window is the smallest ten-thousand boundary that leaves at least one per cent headroom above what this entry measured on the commit named. The rule is stated because the previous one was not a margin at all: the next boundary above the measurement, whatever room that happened to leave, put detection 63 tokens under its declaration and the entry was breached twice in one day. A boundary is a round number; the headroom has to be asked for separately. Measured on the commit named, this entry is over the 200,000-token target and the size report says so on every assembly at this position; the spec's pre-landing estimate of about 121,000 predates itd-194's brief-chapter rows, which add four chapters to every entry naming brief-section, and the figure recorded here is the one measured. Changing any part of this entry is a commit.",
"object": {
"records": [
"itd-177", "itd-178", "itd-179", "itd-180", "itd-181", "itd-182",
"itd-183", "itd-184", "itd-185", "itd-186", "itd-187", "itd-188",
"itd-189", "itd-198", "itd-199",
"itd-194", "itd-2609021003095168", "itd-2609020625400169",
"itd-2609020625400194", "itd-2609020625400445", "itd-2609020625402518",
"itd-2609020625402599", "itd-2609020625405170", "itd-2609020625405251",
"itd-177",
"itd-178",
"itd-179",
"itd-180",
"itd-181",
"itd-182",
"itd-183",
"itd-184",
"itd-185",
"itd-186",
"itd-187",
"itd-188",
"itd-189",
"itd-198",
"itd-199",
"itd-194",
"itd-2609021003095168",
"itd-2609020625400169",
"itd-2609020625400194",
"itd-2609020625400445",
"itd-2609020625402518",
"itd-2609020625402599",
"itd-2609020625405170",
"itd-2609020625405251",
"itd-2609020625407419",
"spc-55", "spc-56", "spc-57", "spc-58", "spc-59", "spc-60", "spc-61",
"spc-62", "spc-63", "spc-64", "spc-65", "spc-66", "spc-67", "spc-68",
"spc-55",
"spc-56",
"spc-57",
"spc-58",
"spc-59",
"spc-60",
"spc-61",
"spc-62",
"spc-63",
"spc-64",
"spc-65",
"spc-66",
"spc-67",
"spc-68",
"spc-69"
],
"paths": [
Expand All @@ -61,21 +124,31 @@
"commands/intent.md"
]
},
"kinds": ["brief-section", "glossary-term", "discipline", "spec", "intent-projection"],
"kinds": [
"brief-section",
"glossary-term",
"discipline",
"spec",
"intent-projection"
],
"window": {
"tokens_est": 300000,
"measured_tokens_est": 288028,
"measured_bytes": 1108908,
"measured_at": "cd5e6e19489ab3be4828296b02f945fd71eab4dc"
"tokens_est": 310000,
"measured_tokens_est": 301560,
"measured_bytes": 1161007,
"measured_at": "5c8300bbc7258a6a72a82264b9546ed7d81b54bf"
}
},
"comparative": {
"comment": "The one entry whose object set is a single record. At this position the include table is the whole account of what the reading sees: every other row withdraws, the disciplines row is narrowed by the assembler to itd-191 before this entry is applied, and the candidate set is DERIVED from the record as the one committed widening run at the target whose items carry no disposition and no admission (adr-2609021016272867). So this entry names the repository material passed BESIDE the candidates, which the readings companion's section 7.2 and its ratified position R3 fix as the declared criteria and nothing else; naming a path or another record here would select nothing, because there is nothing else at this position for it to reach. The kind `candidate` is refused as a preset kind for the same reason: a candidate is selected by the derived run, never by an entry. How the window was measured, stated because it could not be measured the way the other three were: a comparative assembly needs a committed widening run to derive, and this repository holds none, so the figure was taken by dry run over a scratch copy of this tree carrying the comparative channel with a three-item widening run planted in it - 1,312 estimated tokens over 5,053 bytes and 7 items, being the criteria discipline at 693 and six candidate fields at 618. `measured_at` names the commit that scratch tree was taken from. The declaration is the smallest ten-thousand boundary that leaves at least one per cent headroom above that figure — the rule the other three entries state, which at a measurement this small is the first boundary there is — and it is where the headroom for a larger widening run sits: the candidate half is bounded by the run rather than by the tree, which is why the preset-windows eval exempts this position by name and the comparative channel's own eval measures it against the fixture run. It is well under the 200,000-token target. Changing any part of this entry is a commit.",
"object": {
"records": ["itd-191"],
"records": [
"itd-191"
],
"paths": []
},
"kinds": ["discipline"],
"kinds": [
"discipline"
],
"window": {
"tokens_est": 10000,
"measured_tokens_est": 1312,
Expand All @@ -87,11 +160,35 @@
"comment": "The same object set. The kinds are every kind the detection definition reads: its object is the shipped tree read against the claim record, and the tree is code, tests, documentation and configuration alike, with source, test and config items marked unscanned in the manifest. The glossary stands here because the design framework's section 9 lists it among what the input assembler reads for this position and its section 7.2 classes brief/glossary/ as committed vocabulary read cold. The window is the smallest ten-thousand boundary that leaves at least one per cent headroom above what this entry measured on the commit named. The rule is stated because the previous one was not a margin at all: the next boundary above the measurement, whatever room that happened to leave, put detection 63 tokens under its declaration and the entry was breached twice in one day. A boundary is a round number; the headroom has to be asked for separately. The declaration is over the 200,000-token target by a factor of four and inside a reader with a million-token window; the size report states the measured figure on every assembly at this position. Two leaner entries are one commit away, measured here: without test about 539,000 estimated tokens, without source and test about 306,000, both still over the target. Changing any part of this entry is a commit.",
"object": {
"records": [
"itd-177", "itd-178", "itd-179", "itd-180", "itd-181", "itd-182",
"itd-183", "itd-184", "itd-185", "itd-186", "itd-187", "itd-188",
"itd-189", "itd-198", "itd-199",
"spc-55", "spc-56", "spc-57", "spc-58", "spc-59", "spc-60", "spc-61",
"spc-62", "spc-63", "spc-64", "spc-65", "spc-66", "spc-67", "spc-68",
"itd-177",
"itd-178",
"itd-179",
"itd-180",
"itd-181",
"itd-182",
"itd-183",
"itd-184",
"itd-185",
"itd-186",
"itd-187",
"itd-188",
"itd-189",
"itd-198",
"itd-199",
"spc-55",
"spc-56",
"spc-57",
"spc-58",
"spc-59",
"spc-60",
"spc-61",
"spc-62",
"spc-63",
"spc-64",
"spc-65",
"spc-66",
"spc-67",
"spc-68",
"spc-69"
],
"paths": [
Expand All @@ -107,12 +204,22 @@
"commands/intent.md"
]
},
"kinds": ["brief-section", "glossary-term", "discipline", "spec", "intent-projection", "doc", "config", "source", "test"],
"kinds": [
"brief-section",
"glossary-term",
"discipline",
"spec",
"intent-projection",
"doc",
"config",
"source",
"test"
],
"window": {
"tokens_est": 860000,
"measured_tokens_est": 842152,
"measured_bytes": 3242287,
"measured_at": "cd5e6e19489ab3be4828296b02f945fd71eab4dc"
"tokens_est": 910000,
"measured_tokens_est": 899378,
"measured_bytes": 3462609,
"measured_at": "40361877c4006c7eca3e6fd0cfd56a66d0f59fc8"
}
}
}
Expand Down
Loading