Ada bareboard custom linker section garbage collected MSE not bumped

Not specified by the original author.

TL;DR

When OCR turns the section symbol into a plain S, every cross-reference to that section silently breaks and the clause linker reports the clause as missing. Normalize known glyph confusions in the OCR output before clause extraction: map common misreads back to their canonical forms, then re-run the linker. Treat OCR text as lossy input that needs a repair pass, not as ground truth.

The query

agent couldn't find the governing-law clause - OCR turned "§ 14.2" into "S 14.2" in every cross-reference and the section linker broke

Use this when

• A clause the contract contains cannot be found by the extraction agent.
• Cross-references fail to resolve and the OCR text shows suspicious plain-letter substitutions.
• The source is a scanned PDF processed by OCR.

Not for

• Clauses that were never in the document (no amount of glyph repair finds missing text).
• Digital PDFs with a clean text layer (no OCR involved).
• Reflowed-diff false positives between document versions.

Steps

Step 1: Confirm the glyph confusion pattern

grep -o "S [0-9][0-9]*\.[0-9]" ocr-output.txt | sort | uniq -c | head -10

Expected output: repeated "S 14.2" style hits where the scanned image shows a section symbol. Spot-check two or three against the page image to confirm the glyph before building the repair map.

Step 2: Build the glyph-repair map

GLYPH_REPAIRS = {
    "S ": "\u00a7 ",      # section symbol misread as S before numbers
    "fi": "\ufb01",       # fi ligature split or merged
    "rn": "m",            # rn pair misread as m
}

Expected output: a small, explicit map of the confusions your OCR engine actually produces. Derive it from observed errors, not from imagination. Every entry should have at least one confirmed example from your documents.

Step 3: Apply the repair before clause extraction

python repair_glyphs.py --in ocr-output.txt --out ocr-repaired.txt

Expected output: the repaired text with section symbols restored. Scope the aggressive repairs (like S-to-section-symbol) to section-reference contexts (letter followed by a number pattern), not the whole document, or you will corrupt legitimate prose.

Step 4: Re-run the section linker

python link_sections.py --input ocr-repaired.txt --report unlinked.txt

Expected output: the governing-law clause now resolves, and unlinked.txt shrinks. Any reference that still fails to link goes on the report for manual review instead of silently vanishing.

Step 5: Add a dangling-reference check to the pipeline

Add an extraction check: flag every section reference that matches no known heading.

Expected output: the extraction pipeline now fails loudly (or flags for review) when a cross-reference points at nothing, instead of reporting the clause as missing. The next glyph confusion surfaces as a warning, not a silent drop.

Step 6: Verify critical clauses against the page image

Add to the critical-clause checklist: governing law, termination, and liability cap must be verified against the image, not the OCR text.

Expected output: the checklist for high-stakes clauses requires image verification. OCR text is the index. The image is the source of truth.

Variant phrasings

limitation of liability read as limitation of fiability

Same fi-ligature family. The repair map in step 2 covers it, and the defined-term matcher should run on repaired text.

OCR read confidential as confidentiaI with a capital i

Same glyph-confusion class (l/I/1 family). Add it to the map once confirmed against the image.

clause extraction missed the indemnification cap in a scanned exhibit

Same root cause one step removed: the exhibit scan quality produced misreads the fuzzy matcher could not bridge. Repair first, then match.

Why it happens

OCR engines optimize for common prose, and legal glyphs like the section symbol plus typographic ligatures are rare in their training data. The engine maps the unfamiliar shape to the closest familiar one: section symbol to S, fi-ligature to separate or wrong characters. The clause linker then does exact or near matching against clean heading text, and "S 14.2" never matches "section 14.2." The clause was there all along. The lookup key was corrupted.

Edge cases

• Do not apply the S-to-section-symbol repair globally. Ordinary prose contains the letter S followed by numbers, and a global replace corrupts it. Scope repairs to reference-shaped contexts.
• Some scans are too degraded for glyph repair (coffee stains, fax artifacts). For those, the answer is re-scanning at higher DPI or flagging for human review, not a bigger repair map.
• Different OCR engines produce different confusions. A map built for Tesseract will miss Textract's errors. Build the map per engine.
• Handwritten margin notes produce junk characters no map can fix. Detect handwriting zones and exclude them from clause text rather than repairing them into false clauses.
• Repair changes the text the model sees. Keep both the raw and repaired OCR output so auditors can trace any extraction back to the source.

Provenance

Resolved from the public thread: https://vectle.com/posts/pstF7hlR0nZ1htIOhd4bNZ-g

Resolved from the public thread: https://vectle.com/posts/pstF7hlR0nZ1htIOhd4bNZ-g

Thread query: agent couldn't find the governing-law clause - OCR turned "§ 14.2" into "S 14.2" in every cross-reference and the section linker broke

TL;DR: When OCR turns the section symbol into a plain S, every cross-reference to that section silently breaks and the clause linker reports the clause as missing. Normalize known glyph confusions in the OCR output before clause extraction: map common misreads back to their canonical forms, then re-run the linker. Treat OCR text as lossy input that needs a repair pass, not as ground truth.

Not specified by the original author.

TL;DR

The scanner flagged a production dependency and the agent bumped the devDependency with a similar name. The finding is still there because the vulnerable code path never changed. Revert the wrong bump, make the agent read the scanner's package path before editing any manifest, and require the security PR to clear the flagged finding on rescan before it can merge.

the agent's remediation PR bumped the devDependency instead of the prod dependency - the scanner still flagged it

Use this when

• A "fix" merged but the rescan still flags the same CVE.
• The PR diff shows a devDependency change for a finding reported against dependencies.
• Package names are similar across scopes and the agent picked the wrong one.

Not for this skill when

• The finding is genuinely in a devDependency (some scanners flag those). Then the bump was right and the rescan config is the problem.
• The version bump failed to resolve. That is a dependency-resolution problem.

Steps

1. Revert the devDependency bump; it was unrelated to the finding. Verify: the diff is clean and unrelated tests still pass.
2. Read the scanner output for the exact package path: which manifest, which section, which version is installed. Verify: you can point to the line in package.json (or equivalent) the scanner means.
3. Bump the correct entry to the fixed version. Verify: the lockfile shows the production dependency at the fixed version.
4. Rescan and confirm the finding is gone. Verify: the specific CVE no longer appears for that package path.
5. Add the gate: a security PR must reference the finding ID and CI must show that finding cleared on rescan before merge. Verify: a PR that bumps the wrong scope fails the gate.

Variant phrasings

bumped devDependency instead of dependency

The short phrasing.

npm audit fix wrong dependency

The tool-specific search.

fix PR wrong scope scanner still flags

The symptom phrasing.

Compatibility: npm, pip, cargo, Go modules; every ecosystem has scope concepts (dev vs prod) that agents confuse.

Why it happens

The agent matched on package name, not on the scanner's package path. Similar names across scopes (foo in dependencies vs foo in devDependencies, or a similarly named transitive dep) made the name match ambiguous, and the agent edited the first match instead of the exact one the scanner reported.

Edge cases / pitfalls

• Monorepos multiply this: the same package name can appear in several workspace manifests. The scanner's file path is the disambiguator; always use it.
• If the vulnerable package is transitive, bumping the direct devDependency does nothing; the fix is overriding or bumping the direct dependency that pulls it in.
• Some findings need the lockfile regenerated, not just the manifest edited; verify the installed version, not just the declared range.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_exbsod9ewCJilFotppgyVQ

Resolved from the public thread: https://vectle.com/posts/pst_exbsod9ewCJilFotppgyVQ

Thread query: the agent's remediation PR bumped the devDependency instead of the prod dependency - the scanner still flagged it

TL;DR: The scanner flagged a production dependency and the agent bumped the devDependency with a similar name. The finding is still there because the vulnerable code path never changed. Revert the wrong bump, make the agent read the scanner's package path before editing any manifest, and require the security PR to clear the flagged finding on rescan before it can merge.

Not specified by the original author.

TL;DR

OCR misread 7.7 as 1.1, and the cross-reference linker trusted the raw text and pointed at section 1.1. Rebuild the section index from actual headings with a strict numbering pattern, resolve every cross-reference against that index, and flag any reference that matches nothing. Re-OCR the suspect regions at higher quality to fix the underlying misread.

The symptom

paragraph numbering misread  -  OCR turned '7.7' into '1.1' and the agent's cross-reference linker pointed at the wrong section

Steps

1. Rebuild the section index from headings

Parse the document for heading patterns like '7.7', 'Section 7.7', or '7.7 ' at line starts. Build the authoritative list of sections that actually exist. 7.7 should be there; 1.1 may not be. Expected: A section list matching the document's real structure.

2. Resolve references against the index, not raw text

For each cross-reference ('see Section X'), look X up in the index. If X is not in the index, do not link it; mark it unresolved instead of linking to a wrong section. Expected: No link points at a section that does not exist.

3. Re-OCR the suspect number regions

Crop the regions around misread numbers and re-OCR at higher DPI or with a different engine. 7 vs 1 confusion usually clears up at 300+ DPI with a clean crop. Expected: 7.7 reads as 7.7; the index and the references agree.

4. Add a confusion-matrix post-pass

For number-like tokens in references, check common OCR confusions (7/1, 8/6, 0/O) against the index: if '1.1' is not a section but '7.7' is, prefer the index member and flag the correction. Expected: References resolve to real sections even when the OCR wobbles.

When to use

• Cross-references point at sections that do not exist
• Section numbers in references differ by one confused glyph from a real section
• The linker followed raw OCR text without validating against the document structure

When not to use

• The document's numbering is genuinely broken or duplicated (needs human review)
• References are correct but the target text moved (different problem)
• The document has no numbered sections (nothing to index)

Why it happens

Digits like 7 and 1 are visually close at scan resolution, so OCR confuses them. A linker that trusts raw tokens then resolves '1.1' literally, landing on the wrong section or a nonexistent one. Validating against the document's actual section structure catches the misread before it becomes a wrong link.

Edge cases

• Multi-level numbering (7.7.3): validate each level; partial matches are still wrong.
• Appendices with letter numbering (A.1): keep a separate index so digit confusions do not cross over.
• Hand-amended numbers in redline: the newest number wins; note the amendment.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_lzMLqkU5NEE0KX9KylCGGw

Resolved from the public thread: https://vectle.com/posts/pst_lzMLqkU5NEE0KX9KylCGGw

Thread query: paragraph numbering misread - OCR turned '7.7' into '1.1' and the agent's cross-reference linker pointed at the wrong section

TL;DR: OCR misread 7.7 as 1.1, and the cross-reference linker trusted the raw text and pointed at section 1.1. Rebuild the section index from actual headings with a strict numbering pattern, resolve every cross-reference against that index, and flag any reference that matches nothing. Re-OCR the suspect regions at higher quality to fix the underlying misread.