Skip to content

Commit 3eff09c

Browse files
committed
Merge upstream DeusData/codebase-memory-mcp through e783f73 (step 3 of upstream sync)
Brings upstream history from ce0b4b2 (2026-09-16) through e783f73, the first-parent midpoint between step 2 and upstream/main, into the CLI fork. This is the first of two merges that finish the sync; step 4 merges upstream/main itself. 42 files conflicted. Five files the fork deleted were modified upstream: src/mcp/mcp.c, src/mcp/mcp_internal.h, tests/test_mcp.c, src/cli/client_adapter.c and server.json. They stay deleted, and the upstream behaviour in them is moved to the CLI's operation layer. Moved from upstream MCP handlers into src/operations (CLI commands): - index resource limits (config keys index_max_files and index_max_source_mb, docs/INDEX_RESOURCE_LIMITS.md): * The index operation loads the operator's policy from _config.db and encodes it on the worker request as _cbm_index_policy. Any caller-supplied field with that name is stripped first, so a request cannot relax a guard. * A worker trusts only the policy its supervisor encoded and rejects a request without one ("missing or incomplete trusted worker policy"). * Discovery stops at the first file or byte over the limit. The response is code=resource_limit_exceeded with stage, resource, observed, limit, unit, retryable and serving_index_preserved. The previous database is never replaced by a partial index. * src/daemon/application.c adds the policy to auto-index jobs and to programmatic index requests. cbm_index_operation_policy_add_to_args() in operations/index.c replaces upstream's cbm_mcp_index_policy_add_to_args(). - search_graph query mode (BM25), in operations/search.c: * The label filter now applies. It was silently ignored, so label=Class returned Methods. The count query mirrors the filter. * Exact-name tier: a definition whose name equals the query ranks above partial hits, then a case-insensitive exact match, then BM25 order. - check_index_coverage freshness: mtime_ns is read through cbm_path_info_utf8, the source the indexer records with. Recomputing it from struct stat truncated to seconds on Windows, so unchanged files were reported metadata_changed (DeusData#1714). - Merged without a port needed: * `config set/reset` validation for the two limit keys. * `--paths '["a","b"]'` accepted as a JSON array literal in cli_add_typed. * discover and pipeline enforcement of the limits. * Waste-sanitizer memory events (src/foundation/mem_events.c, MEMWASTE=1 Makefile flavour, scripts/memwaste*.py), the arena-backed walk_defs frames and the new language mappings (.resx, .xaml, MSBuild files). Dropped as MCP-protocol-only: - src/cli/client_adapter.c, including the Pi TypeBox schema generator. - server.json and the codebase-memory-mcp package names (the fork keeps its codebase-memory-cli names; version bumps to 0.11.0 are merged). - MCP registration expectations in the Codex install tests (test_cli.c). The fork's instruction-block-only tests stay. - The Pi/OpenCode adapter tests in test_agent_clients.c. - memlab-drive.py JSON-RPC project scoping and isError handling. The CLI driver has no JSON-RPC session. - The soak harness's stdio server flow. - The MCP marketing text in README.md, docs/index.html and pkg/npm/README.md. Notable resolutions: - Same fix on both sides, so upstream's wording was taken: * watcher.c and watcher.h hard-failure index backoff (the fork's version was a hand-port; test_watcher.c keeps the fork's index_backoff_is_bounded_and_monotonic beside upstream's six tests). * go_lsp.c parse_field_defs split. * language.c comments. * The cbm_utf8_trim_partial and cbm_is_project_index_db halves of str_util.c are both kept. * test_extraction.c keeps one copy of the DeusData#2210 await test. - Genuine fork behaviour kept: * extract_defs.c walk_defs: the fork's no-ceiling growth with a per-file error on allocation failure, plus upstream's arena-backed frames. * Codex AGENTS.md block tests, the fork's README, CLI names in package metadata, and the CLI-first smoke-test.sh and memlab-drive.py. * main.c worker exit path: cbm_memev_process_exit() runs before the wire response is encoded. - Duplicates git kept twice, now removed: * unwrap_await_callee in extract_calls.c: upstream's ungated copy is kept. The fork's JS/TS-gated copy is dropped, which is a small behaviour change. * unwrap_await_callee in ts_lsp.c: the earlier definition is kept because the fork uses it before upstream's position. * binding_has_value in cypher.c: the earlier definition is kept for the WITH COUNT path. * cbm_line_offsets in cbm.c: upstream's cbm_alloc version, with the fork's macro-line scan counter at the call site. * The update-notice thread block in application.c, which the fork deleted. * mem_override_posix.c and mem_profile.c in FOUNDATION_SRCS. Upstream replaced mem_profile with mem_events, and the POSIX wrap now links only under MEMWASTE=1. - soak-test.sh called generate_project without defining it, since step 2 or earlier. The definition is restored from upstream. - tests/test_index_policy.c now drives the operation host instead of cbm_mcp_handle_tool. The MCP text-wrapped responses are replaced by the raw JSON the CLI returns. - New CLI-layer tests in test_integration.c replace the dropped test_mcp.c cases: BM25 label filter, BM25 exact-name ranking and DeusData#1714 coverage freshness. - Jenkinsfile suite list gains index_policy and mem_events. - Vendored sources match upstream except rescript/scanner.c, which is back at upstream's e783f73 bytes; the fork's DeusData#2176 port of it returns with step 4. All vendored checksums verify. Known gaps: - smoke_dump_worker_log (print the index worker log when a smoke run fails) is not ported into scripts/smoke-invariants.sh. - The first-failure line in memlab-drive.py was not ported. The production build and the test-runner compile are clean with -Werror. The index_policy and integration suites pass locally; the full suites run on Jenkins.
2 parents 8c52d7a + e783f73 commit 3eff09c

168 files changed

Lines changed: 16394 additions & 1700 deletions

File tree

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

‎.github/workflows/_memwaste.yml‎

Lines changed: 183 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,183 @@
1+
# Reusable: the waste sanitizer (memory + CPU waste), REPORT-ONLY.
2+
#
3+
# What runs (scripts/memwaste.sh, memwaste-scaling.py, memwaste-ratchet.py):
4+
# mode=pr Linux: the event lane (memory waste + CPU work) indexing this
5+
# repository at the checked-out commit, and the CPU scaling lane
6+
# (src/ replicated 2x against 4x, per-function counter ratios).
7+
# mode=full the same, plus the event lane on macOS and Windows and the access
8+
# lane (read/write verdicts per allocation) on Linux.
9+
#
10+
# The corpus is the repository itself: pinned by the commit, no network, and a
11+
# real mix of C, C++, Python, JavaScript and vendored code.
12+
#
13+
# REPORT-ONLY: every job uploads its report and prints the ratchet comparison
14+
# against scripts/memwaste-baseline.json, but nothing here can turn a PR red --
15+
# the ratchet runs without --enforce and the PR's ci-ok gate does not depend on
16+
# these jobs. A job fails only when the MEASUREMENT is unsound (a table
17+
# overflowed, frees the layer never saw): that is a defect of the tool, and a
18+
# red job is how it gets noticed. Switching the ratchet to --enforce is a
19+
# separate, deliberate change once the numbers have been stable across reruns.
20+
name: Memwaste
21+
22+
on:
23+
workflow_call:
24+
inputs:
25+
mode:
26+
description: "pr (Linux event + scaling) or full (all platforms, access lane)"
27+
type: string
28+
default: pr
29+
30+
permissions:
31+
contents: read
32+
33+
jobs:
34+
event-linux:
35+
runs-on: ubuntu-latest
36+
timeout-minutes: 60
37+
steps:
38+
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
39+
- name: Install deps
40+
run: sudo apt-get update && sudo apt-get install -y zlib1g-dev python3 clang llvm
41+
- name: Event lane on this repository
42+
run: |
43+
# `cmd | tee` reports TEE's status: without this a failing lane
44+
# (a missing exec bit, a build error) passes the step and only
45+
# surfaces as the ratchet not finding its JSON.
46+
set -o pipefail
47+
mkdir -p "$RUNNER_TEMP/memwaste-event"
48+
scripts/memwaste.sh "$GITHUB_WORKSPACE" --top 40 --out "$RUNNER_TEMP/memwaste-event" \
49+
| tee "$RUNNER_TEMP/memwaste-event/report.txt"
50+
- name: Ratchet (report-only)
51+
run: python3 scripts/memwaste-ratchet.py --out "$RUNNER_TEMP/memwaste-event" --venue ${{ inputs.mode }}-linux
52+
- name: Upload report
53+
if: ${{ always() }}
54+
uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
55+
with:
56+
name: memwaste-event-linux-${{ inputs.mode }}
57+
path: |
58+
${{ runner.temp }}/memwaste-event/report.txt
59+
${{ runner.temp }}/memwaste-event/memwaste.json
60+
${{ runner.temp }}/memwaste-event/index.json
61+
if-no-files-found: warn
62+
63+
scaling-linux:
64+
runs-on: ubuntu-latest
65+
timeout-minutes: 60
66+
steps:
67+
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
68+
- name: Install deps
69+
run: sudo apt-get update && sudo apt-get install -y zlib1g-dev python3 clang llvm
70+
- name: CPU scaling lane (src/ at 2x against 4x)
71+
# Report-only: the script exits 1 when it flags a super-linear function.
72+
continue-on-error: true
73+
run: |
74+
# `cmd | tee` reports TEE's status: without this a failing lane
75+
# (a missing exec bit, a build error) passes the step and only
76+
# surfaces as the ratchet not finding its JSON.
77+
set -o pipefail
78+
mkdir -p "$RUNNER_TEMP/memwaste-scaling"
79+
python3 scripts/memwaste-scaling.py --corpus src --replicas 2 \
80+
--out "$RUNNER_TEMP/memwaste-scaling" | tee "$RUNNER_TEMP/memwaste-scaling/report.txt"
81+
- name: Upload report
82+
if: ${{ always() }}
83+
uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
84+
with:
85+
name: memwaste-scaling-linux-${{ inputs.mode }}
86+
path: ${{ runner.temp }}/memwaste-scaling/report.txt
87+
if-no-files-found: warn
88+
89+
access-linux:
90+
if: ${{ inputs.mode == 'full' }}
91+
runs-on: ubuntu-latest
92+
timeout-minutes: 120
93+
steps:
94+
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
95+
- name: Install deps
96+
run: sudo apt-get update && sudo apt-get install -y zlib1g-dev python3 clang llvm
97+
- name: Access lane on this repository
98+
run: |
99+
# `cmd | tee` reports TEE's status: without this a failing lane
100+
# (a missing exec bit, a build error) passes the step and only
101+
# surfaces as the ratchet not finding its JSON.
102+
set -o pipefail
103+
mkdir -p "$RUNNER_TEMP/memwaste-access"
104+
scripts/memwaste.sh "$GITHUB_WORKSPACE" --lane access --top 40 --sort untouched \
105+
--out "$RUNNER_TEMP/memwaste-access" | tee "$RUNNER_TEMP/memwaste-access/report.txt"
106+
- name: Upload report
107+
if: ${{ always() }}
108+
uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
109+
with:
110+
name: memwaste-access-linux-${{ inputs.mode }}
111+
path: |
112+
${{ runner.temp }}/memwaste-access/report.txt
113+
${{ runner.temp }}/memwaste-access/memwaste.json
114+
if-no-files-found: warn
115+
116+
event-macos:
117+
if: ${{ inputs.mode == 'full' }}
118+
runs-on: macos-14
119+
timeout-minutes: 60
120+
steps:
121+
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
122+
- name: Event lane on this repository
123+
run: |
124+
# `cmd | tee` reports TEE's status: without this a failing lane
125+
# (a missing exec bit, a build error) passes the step and only
126+
# surfaces as the ratchet not finding its JSON.
127+
set -o pipefail
128+
mkdir -p "$RUNNER_TEMP/memwaste-event"
129+
scripts/memwaste.sh "$GITHUB_WORKSPACE" --top 40 --out "$RUNNER_TEMP/memwaste-event" \
130+
| tee "$RUNNER_TEMP/memwaste-event/report.txt"
131+
- name: Ratchet (report-only)
132+
run: python3 scripts/memwaste-ratchet.py --out "$RUNNER_TEMP/memwaste-event" --venue full-macos
133+
- name: Upload report
134+
if: ${{ always() }}
135+
uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
136+
with:
137+
name: memwaste-event-macos-${{ inputs.mode }}
138+
path: |
139+
${{ runner.temp }}/memwaste-event/report.txt
140+
${{ runner.temp }}/memwaste-event/memwaste.json
141+
if-no-files-found: warn
142+
143+
event-windows:
144+
if: ${{ inputs.mode == 'full' }}
145+
runs-on: windows-latest
146+
timeout-minutes: 90
147+
steps:
148+
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
149+
- uses: msys2/setup-msys2@66cd2cce69caa17b53920067426061ca1de3a884 # v2
150+
with:
151+
msystem: CLANG64
152+
path-type: inherit
153+
install: >-
154+
mingw-w64-clang-x86_64-clang
155+
mingw-w64-clang-x86_64-llvm
156+
mingw-w64-clang-x86_64-zlib
157+
mingw-w64-clang-x86_64-python3
158+
make
159+
git
160+
coreutils
161+
- name: Event lane on this repository
162+
shell: msys2 {0}
163+
run: |
164+
# `cmd | tee` reports TEE's status: without this a failing lane
165+
# (a missing exec bit, a build error) passes the step and only
166+
# surfaces as the ratchet not finding its JSON.
167+
set -o pipefail
168+
out="$(cygpath -u "$RUNNER_TEMP")/memwaste-event"
169+
mkdir -p "$out"
170+
CC=clang CXX=clang++ scripts/memwaste.sh "$(cygpath -u "$GITHUB_WORKSPACE")" --top 40 --out "$out" \
171+
| tee "$out/report.txt"
172+
- name: Ratchet (report-only)
173+
shell: msys2 {0}
174+
run: python3 scripts/memwaste-ratchet.py --out "$(cygpath -u "$RUNNER_TEMP")/memwaste-event" --venue full-windows
175+
- name: Upload report
176+
if: ${{ always() }}
177+
uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
178+
with:
179+
name: memwaste-event-windows-${{ inputs.mode }}
180+
path: |
181+
${{ runner.temp }}/memwaste-event/report.txt
182+
${{ runner.temp }}/memwaste-event/memwaste.json
183+
if-no-files-found: warn

‎.github/workflows/codeql.yml‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -29,7 +29,7 @@ jobs:
2929
run: sudo apt-get update && sudo apt-get install -y zlib1g-dev
3030

3131
- name: Initialize CodeQL
32-
uses: github/codeql-action/init@cdf488f595d80d6e07e03d4674febd5ab45fa938 # v4.37.9
32+
uses: github/codeql-action/init@b96794f015dfd88f77b49b1c93e0fa7110f94c63 # v4.38.0
3333
with:
3434
languages: c-cpp
3535
build-mode: manual
@@ -38,6 +38,6 @@ jobs:
3838
run: scripts/build.sh
3939

4040
- name: Perform CodeQL Analysis
41-
uses: github/codeql-action/analyze@cdf488f595d80d6e07e03d4674febd5ab45fa938 # v4.37.9
41+
uses: github/codeql-action/analyze@b96794f015dfd88f77b49b1c93e0fa7110f94c63 # v4.38.0
4242
with:
4343
category: "/language:c-cpp"

‎.github/workflows/dry-run.yml‎

Lines changed: 36 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -45,6 +45,10 @@ on:
4545
description: 'Skip candidate scan; mark unscanned and default to stripped'
4646
type: boolean
4747
default: false
48+
skip_memwaste:
49+
description: 'Skip the waste sanitizer (report-only) and the fuzz smoke'
50+
type: boolean
51+
default: false
4852
soak_level:
4953
description: 'Soak: full (quick+asan), quick (10min), none'
5054
type: choice
@@ -107,6 +111,38 @@ jobs:
107111
with:
108112
broad_platforms: true
109113

114+
# ── Waste sanitizer, REPORT-ONLY (independent island) ─────────
115+
# Event lane on Linux, macOS and Windows, access lane and the per-function
116+
# scaling lane on Linux. Uploads reports; fails only when the measurement is
117+
# unsound (a defect of the tool itself).
118+
memwaste:
119+
if: ${{ inputs.skip_memwaste != true }}
120+
uses: ./.github/workflows/_memwaste.yml
121+
with:
122+
mode: full
123+
124+
# ── Fuzz smoke (independent island) ───────────────────────────
125+
# A fixed number of executions from a fixed seed per target, so the verdict is
126+
# a function of the commit (O9). A crash here is a real defect in a parser
127+
# that reads untrusted input; its reproducer is uploaded.
128+
fuzz:
129+
if: ${{ inputs.skip_memwaste != true }}
130+
runs-on: ubuntu-latest
131+
timeout-minutes: 90
132+
steps:
133+
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
134+
- name: Install deps
135+
run: sudo apt-get update && sudo apt-get install -y zlib1g-dev clang llvm libclang-rt-dev
136+
- name: Fuzz smoke (extract, cypher, config)
137+
run: scripts/fuzz.sh all --runs 20000 --seed 1
138+
- name: Upload reproducers
139+
if: ${{ failure() }}
140+
uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
141+
with:
142+
name: fuzz-artifacts
143+
path: build/fuzz/artifacts-*/
144+
if-no-files-found: ignore
145+
110146
# ── Soak tests (optional, parallel with smoke) ────────────────
111147
soak:
112148
if: ${{ inputs.soak_level != 'none' && !cancelled() && needs.build.result != 'failure' && needs.build.result != 'skipped' }}

‎.github/workflows/nightly-soak.yml‎

Lines changed: 21 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -35,6 +35,27 @@ permissions:
3535
contents: read
3636

3737
jobs:
38+
# Fuzz exploration: new inputs are the point here, so it runs for a time
39+
# budget instead of a fixed count. NON-GATING like the soak; a crash uploads
40+
# its reproducer, which becomes a fixed-seed regression input once fixed.
41+
fuzz:
42+
if: ${{ github.event_name == 'schedule' || github.event_name == 'workflow_dispatch' }}
43+
runs-on: ubuntu-latest
44+
timeout-minutes: 120
45+
steps:
46+
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
47+
- name: Install deps
48+
run: sudo apt-get update && sudo apt-get install -y zlib1g-dev clang llvm libclang-rt-dev
49+
- name: Fuzz exploration (30 min per target)
50+
run: scripts/fuzz.sh all --max-total-time 1800 --seed 0
51+
- name: Upload reproducers
52+
if: ${{ failure() }}
53+
uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
54+
with:
55+
name: fuzz-artifacts-nightly
56+
path: build/fuzz/artifacts-*/
57+
if-no-files-found: ignore
58+
3859
soak:
3960
uses: ./.github/workflows/_soak.yml
4061
with:

‎.github/workflows/pr.yml‎

Lines changed: 33 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -64,9 +64,29 @@ jobs:
6464
run: |
6565
# The full .diff endpoint rejects large-but-valid PRs at 20k lines.
6666
# The paginated files endpoint remains filename-only for this gate.
67-
FILES=$(gh api --paginate "repos/$REPO/pulls/$PR/files?per_page=100" --jq '.[].filename')
67+
if FILES=$(gh api --paginate "repos/$REPO/pulls/$PR/files?per_page=100" --jq '.[].filename'); then
68+
echo "file list: pulls/files API"
69+
else
70+
# The files endpoint renders the diff server-side too, and answers
71+
# 422 "this diff is taking too long to generate" for a PR that
72+
# regenerates a vendored parser (#2246: a 42 MB sql/parser.c) --
73+
# so does the compare endpoint. The PR merge ref's first parent is
74+
# the base, so a name-only diff across it is the same file list.
75+
# depth 2 + blob:none fetches commits and trees only (well under
76+
# 1 MB, under a second); nothing from the PR is checked out or run.
77+
echo "::notice::pulls/files API failed; file list taken from refs/pull/$PR/merge"
78+
CHANGES_GIT="$RUNNER_TEMP/changes.git"
79+
git init -q --bare "$CHANGES_GIT"
80+
git --git-dir="$CHANGES_GIT" fetch -q --depth=2 --filter=blob:none \
81+
"https://github.com/$REPO" "refs/pull/$PR/merge"
82+
FILES=$(git --git-dir="$CHANGES_GIT" diff --name-only --no-renames FETCH_HEAD^1 FETCH_HEAD)
83+
fi
6884
printf '%s\n' "$FILES"
69-
if printf '%s\n' "$FILES" | grep -qE '^(src/|internal/|install\.(sh|ps1)|scripts/build\.sh|scripts/smoke-test\.sh|scripts/smoke-local\.sh|scripts/smoke-fixture-server\.py|scripts/gen-third-party-notices\.sh|scripts/env\.sh|test-infrastructure/vm/(vm-smoke\.sh|windows-user-path-guard\.ps1)|Makefile\.cbm)'; then
85+
# A here-string, not a pipe: under this shell's pipefail, grep -q
86+
# exits on its first match, printf dies of SIGPIPE once the list
87+
# outgrows the pipe buffer, and the MATCH is then reported as a
88+
# failure -- product=false, smoke skipped, on the largest PRs.
89+
if grep -qE '^(src/|internal/|install\.(sh|ps1)|scripts/build\.sh|scripts/smoke-test\.sh|scripts/smoke-local\.sh|scripts/smoke-fixture-server\.py|scripts/gen-third-party-notices\.sh|scripts/env\.sh|test-infrastructure/vm/(vm-smoke\.sh|windows-user-path-guard\.ps1)|Makefile\.cbm)' <<<"$FILES"; then
7090
echo "product=true" >> "$GITHUB_OUTPUT"
7191
else
7292
echo "product=false" >> "$GITHUB_OUTPUT"
@@ -154,6 +174,17 @@ jobs:
154174
CCACHE_DIR: ${{ github.workspace }}/.ccache
155175
CCACHE_MAXSIZE: 1000M
156176

177+
# ── Waste sanitizer, REPORT-ONLY: memory + CPU waste on this repository and
178+
# the per-function scaling lane (Linux). Deliberately NOT in ci-ok's needs:
179+
# it uploads evidence and prints the ratchet, it never blocks a merge. Only
180+
# product-code PRs pay for it. ──
181+
memwaste:
182+
needs: [changes]
183+
if: ${{ !cancelled() && needs.changes.outputs.product == 'true' }}
184+
uses: ./.github/workflows/_memwaste.yml
185+
with:
186+
mode: pr
187+
157188
ci-ok:
158189
# The one required context (besides dco) — fails unless every PR stage
159190
# succeeded, so matrix renames can never silently deadlock merges. `skipped`

‎.github/workflows/release.yml‎

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -191,6 +191,15 @@ jobs:
191191
use_release_artifacts: true
192192
platforms: ${{ inputs.platforms }}
193193

194+
# ── Waste sanitizer, REPORT-ONLY ────────────────────────────────
195+
# The same full run as the dry run, on the release commit. Not in
196+
# release-draft's needs: the report is evidence for the release notes, never
197+
# a publication gate.
198+
memwaste:
199+
uses: ./.github/workflows/_memwaste.yml
200+
with:
201+
mode: full
202+
194203
# ── 6. Create DRAFT release ────────────────────────────────────
195204
# Requires smoke to have actually SUCCEEDED, not merely "not failed". The bare
196205
# !cancelled() && !failure() form is fail-OPEN: a skipped smoke is neither

‎.github/workflows/scorecard.yml‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -27,6 +27,6 @@ jobs:
2727
publish_results: true
2828

2929
- name: Upload SARIF results
30-
uses: github/codeql-action/upload-sarif@cdf488f595d80d6e07e03d4674febd5ab45fa938 # v4
30+
uses: github/codeql-action/upload-sarif@b96794f015dfd88f77b49b1c93e0fa7110f94c63 # v4
3131
with:
3232
sarif_file: results.sarif

‎CONTRIBUTING.md‎

Lines changed: 19 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -149,6 +149,25 @@ If in doubt, open an issue and ask.
149149
- Run `scripts/test.sh` and `scripts/lint.sh` before submitting
150150
- Keep PRs focused — avoid unrelated reformatting or refactoring
151151

152+
## AI-assisted and AI-authored contributions
153+
154+
Contributions written with the help of an AI tool, or written by an AI agent acting for a person, are welcome here on the same terms as any other. What we review is the change, not who or what typed it. Three things make that work:
155+
156+
- **Say so.** If an agent wrote the PR or is answering in the thread, disclose it — a line in the PR description or at the top of a comment is enough ("written with X", "this account is operated by an AI agent on behalf of Y"). Disclosure is never held against a contribution. Finding out later is.
157+
- **A person is accountable.** The [DCO sign-off](#license-and-sign-off-dco--required-on-every-commit) is a legal certification, and only a person can make it. The human behind the account certifies every commit, answers for it, and is who we talk to if something goes wrong. An agent may carry the conversation; it cannot carry the responsibility.
158+
- **The bar does not move.** Every claim in a PR must be something a reviewer can check: the failing test, the command that was run, the numbers and the machine they came from. "The model says it works" is not evidence, and a confident description does not replace a test that fails without the fix. We read AI-written PRs exactly as carefully as human-written ones — no more suspiciously, and no less.
159+
160+
Two practical notes:
161+
162+
- `Co-authored-by:` trailers naming a tool are fine. Links to private chat or agent sessions are not — they are dead to everyone else and do not belong in permanent history; we will ask you to remove them.
163+
- High-volume, low-effort submissions — many near-identical PRs, changes nobody ran, replies that do not engage with the review — will be closed, whoever or whatever produced them. One careful PR is worth more than twenty generated ones.
164+
165+
### Who speaks for the project
166+
167+
Only the maintainer account ([@DeusData](https://github.com/DeusData)) speaks for this project. A review, an approval, a "this will be merged" or a "this is a duplicate" from any other account is that person's opinion, however official it sounds — helpful reviews from the community are very welcome, and that is what they are.
168+
169+
**Impersonating a maintainer is not acceptable**, by a person or by an agent: presenting yourself as part of the maintainer team, posting approvals or change requests styled as maintainer decisions, or telling other contributors what the project has decided. We dismiss such reviews, hide the comments, say publicly in the affected threads that they were not ours, and report and block the account. If you are unsure whether a response you received is from the project, check the account name — and ask.
170+
152171
## Security
153172

154173
We take security seriously. All PRs go through:

0 commit comments

Comments
 (0)