Skills extract-transcripts

📦

extract-transcripts

Name: extract-transcripts
Author: 0xBigBoss

Content revision r1 High Risk ⚙️ External commands📁 Filesystem access

Extract AI Session Transcripts

AI coding session histories are difficult to review in raw JSONL form. This skill converts Claude Code and Codex sessions into readable markdown and searchable indexes.

Supports: Claude Codex Code(CC)

⚠️ 38 Poor

Stop for confirmation before installing.

Review the plan and obtain explicit user consent before changing files.

Install with my Agent

Copy this request to your Agent. It includes the canonical Skill page and manifest.

Agent request

Review the Skillstore skill "extract-transcripts" from https://skillstore.io/skills/0xbigboss-extract-transcripts.md and its manifest at https://skillstore.io/api/skills/0xbigboss-extract-transcripts/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.

Your Agent should still show its plan and request any confirmation required by the security policy.

Agent-readable resources

Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.

Markdown detail GET /skills/0xbigboss-extract-transcripts.md Signed manifest GET /api/skills/0xbigboss-extract-transcripts/manifest Signed lockfile GET /api/skills/0xbigboss-extract-transcripts/lockfile

Test it

Using "extract-transcripts". A Claude Code session JSONL file from a local project.

Expected outcome:

A markdown transcript with session date, duration, model, working directory, user messages, and assistant responses.

Using "extract-transcripts". A Codex CLI history file containing several prompts.

Expected outcome:

A dated list of Codex history entries with long prompts shortened for easier scanning.

Using "extract-transcripts". A transcript index search for error handling in one project.

Expected outcome:

Recent matching sessions grouped by date, project, branch, file path, and matching message snippet.

Security Audit

High Risk

v6 • 7/4/2026 Open versioned report

Most static alerts for markdown backticks, option names, and schema identifiers are false positives. Confirmed risks are local command execution instructions, hidden Claude and Codex transcript access, persistent DuckDB indexing of transcript content, and user-selected output writes. No prompt injection or network exfiltration was found in the reviewed files.

Files scanned

1,413

Lines analyzed

Review items

False positives ignored

Confirmed security concerns (3)

High

transcript_index.py:20-25 transcript_index.py:167-175 transcript_index.py:281-289

Persistent Index Stores Full Transcript Content

The indexer stores local session message content in a persistent DuckDB database without redaction or encryption. Claude and Codex transcripts may contain secrets, prompts, tool output, project paths, and repository metadata.

The code defines hidden default session and database paths, extracts message content, and inserts that content into the messages table. This directly supports persistent local storage of sensitive transcript text.

Medium

transcript_index.py:582

SQLite database file

db_path = Path(args.db) if args.db else DEFAULT_DB_PATH

The code chooses DEFAULT_DB_PATH when --db is omitted and connects to that DuckDB file. It stores searchable transcript data locally, so the database path is privacy-sensitive.

Medium

SKILL.md:34-36 extract_transcript.py:182-198

Detailed Exports Can Reveal Tool Inputs And Thinking

The Claude extractor can include tool inputs and thinking blocks in generated markdown. Those fields can expose sensitive commands, arguments, file paths, or reasoning text when outputs are shared.

The documentation advertises include-tools and include-thinking, and the code writes thinking blocks plus serialized tool input into markdown. Exposure depends on user flags, so the risk is medium rather than critical.

Capability review items (39)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

High

PLAN.md:159

Hidden file in home directory

~/.claude/transcript-index/

The plan documents a hidden home directory for a transcript DuckDB index. That location would persist derived session data under ~/.claude and is a real local privacy risk.

High

SKILL.md:50

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/extract_codex_transcript.py ~/.codex/history.jsonl --hi

The command explicitly reads ~/.codex/history.jsonl, a hidden home file containing Codex prompt history. That is intended behavior but exposes sensitive local transcript data.

High

SKILL.md:56

Hidden file in home directory

- Sessions: `~/.claude/projects/<project-path>/<session-id>.jsonl`

The documentation points to hidden Claude or Codex session and history files under the home directory. These files can contain sensitive prompts, tool output, and project metadata.

High

SKILL.md:59

Hidden file in home directory

- Sessions: `~/.codex/sessions/<session_id>/rollout.jsonl`

The documentation points to hidden Claude or Codex session and history files under the home directory. These files can contain sensitive prompts, tool output, and project metadata.

High

SKILL.md:60

Hidden file in home directory

- History: `~/.codex/history.jsonl`

The documentation points to hidden Claude or Codex session and history files under the home directory. These files can contain sensitive prompts, tool output, and project metadata.

High

SKILL.md:68

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/transcript_index.py index

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

High

SKILL.md:71

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/transcript_index.py index --full

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

High

SKILL.md:74

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/transcript_index.py index --limit 10

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

High

SKILL.md:77

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/transcript_index.py recent

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

High

SKILL.md:78

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/transcript_index.py recent --limit 20

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

High

SKILL.md:79

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/transcript_index.py recent --project myapp

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

High

SKILL.md:80

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/transcript_index.py recent --since 7d

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

High

SKILL.md:83

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/transcript_index.py search "error handling"

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

High

SKILL.md:84

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/transcript_index.py search "query" --cwd ~/myproject

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

High

SKILL.md:87

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/transcript_index.py show <file_path>

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

High

SKILL.md:88

Hidden file in home directory

python3 ~/.claude/skills/extract-transcripts/transcript_index.py show <file_path> --summary

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

High

SKILL.md:93

Hidden file in home directory

**Database location:** `~/.claude/transcript-index/sessions.duckdb`

The documentation identifies ~/.claude/transcript-index/sessions.duckdb as the default persistent transcript database. A hidden local database of conversations is a real privacy-sensitive file.

Medium

extract_transcript.py:266

Python file write/append

output_file = open(args.output, 'w')

Opening args.output with write mode allows the tool to create or overwrite a user-supplied path. It is intended output behavior, but it is a real filesystem write capability.

Medium

PLAN.md:159

Hidden file access

~/.claude/transcript-index/

The plan documents a hidden home directory for a transcript DuckDB index. That location would persist derived session data under ~/.claude and is a real local privacy risk.

Medium

SKILL.md:14-32

Ruby/shell backtick execution

```bash

The bash block instructs users or agents to run python3 scripts from this community skill. That is intentional command execution and requires trust in the script contents.

Medium

SKILL.md:45-51

Ruby/shell backtick execution

```bash

The bash block instructs running the Codex extractor and includes a command reading ~/.codex/history.jsonl. This is local command execution over potentially sensitive history.

Medium

SKILL.md:66-89

Ruby/shell backtick execution

```bash

The bash block instructs running the transcript indexer, whose defaults scan local Claude session stores and open a persistent DuckDB database. This is real local command execution.

Medium

SKILL.md:91-93

Ruby/shell backtick execution

**Requirements:** DuckDB (`pip install duckdb`)

The requirement tells users to install duckdb with pip. It is not automatic, but it is an external package install instruction from community documentation.

Medium

SKILL.md:50

Hidden file access

python3 ~/.claude/skills/extract-transcripts/extract_codex_transcript.py ~/.codex/history.jsonl --hi

The command explicitly reads ~/.codex/history.jsonl, a hidden home file containing Codex prompt history. That is intended behavior but exposes sensitive local transcript data.

Medium

SKILL.md:56

Hidden file access

- Sessions: `~/.claude/projects/<project-path>/<session-id>.jsonl`

The documentation points to hidden Claude or Codex session and history files under the home directory. These files can contain sensitive prompts, tool output, and project metadata.

Medium

SKILL.md:59

Hidden file access

- Sessions: `~/.codex/sessions/<session_id>/rollout.jsonl`

The documentation points to hidden Claude or Codex session and history files under the home directory. These files can contain sensitive prompts, tool output, and project metadata.

Medium

SKILL.md:60

Hidden file access

- History: `~/.codex/history.jsonl`

The documentation points to hidden Claude or Codex session and history files under the home directory. These files can contain sensitive prompts, tool output, and project metadata.

Medium

SKILL.md:68

Hidden file access

python3 ~/.claude/skills/extract-transcripts/transcript_index.py index

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

Medium

SKILL.md:71

Hidden file access

python3 ~/.claude/skills/extract-transcripts/transcript_index.py index --full

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

Medium

SKILL.md:74

Hidden file access

python3 ~/.claude/skills/extract-transcripts/transcript_index.py index --limit 10

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

Medium

SKILL.md:77

Hidden file access

python3 ~/.claude/skills/extract-transcripts/transcript_index.py recent

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

Medium

SKILL.md:78

Hidden file access

python3 ~/.claude/skills/extract-transcripts/transcript_index.py recent --limit 20

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

Medium

SKILL.md:79

Hidden file access

python3 ~/.claude/skills/extract-transcripts/transcript_index.py recent --project myapp

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

Medium

SKILL.md:80

Hidden file access

python3 ~/.claude/skills/extract-transcripts/transcript_index.py recent --since 7d

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

Medium

SKILL.md:83

Hidden file access

python3 ~/.claude/skills/extract-transcripts/transcript_index.py search "error handling"

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

Medium

SKILL.md:84

Hidden file access

python3 ~/.claude/skills/extract-transcripts/transcript_index.py search "query" --cwd ~/myproject

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

Medium

SKILL.md:87

Hidden file access

python3 ~/.claude/skills/extract-transcripts/transcript_index.py show <file_path>

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

Medium

SKILL.md:88

Hidden file access

python3 ~/.claude/skills/extract-transcripts/transcript_index.py show <file_path> --summary

This command invokes transcript_index.py, which uses a default hidden database and indexes local transcript content. That is real hidden-file access tied to sensitive session data.

Medium

SKILL.md:93

Hidden file access

**Database location:** `~/.claude/transcript-index/sessions.duckdb`

The documentation identifies ~/.claude/transcript-index/sessions.duckdb as the default persistent transcript database. A hidden local database of conversations is a real privacy-sensitive file.

Risk Factors

⚙️ External commands (22)

📁 Filesystem access (49)

Audited by: codex View Audit History →

Share & cite this report

Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.

Open versioned report

Copy report link

https://skillstore.io/skills/0xbigboss-extract-transcripts/audits/6?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown badge

[![Skillstore security assessment](https://skillstore.io/badges/skills/0xbigboss-extract-transcripts/security.svg)](https://skillstore.io/skills/0xbigboss-extract-transcripts?utm_source=security_passport_badge)

HTML badge

<a href="https://skillstore.io/skills/0xbigboss-extract-transcripts?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/0xbigboss-extract-transcripts/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Embed card

<iframe src="https://skillstore.io/embed/skills/0xbigboss-extract-transcripts.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

Academic citations (APA · BibTeX · CFF)

APA citation

0xBigBoss. (2026). extract-transcripts security audit report (audit version 6) [Author version unspecified]. Skillstore. https://skillstore.io/skills/0xbigboss-extract-transcripts/audits/6

BibTeX citation

@techreport{0xbigboss-0xbigboss-extract-transcripts-2026,
  author = {0xBigBoss},
  title = {extract-transcripts security audit report (audit version 6)},
  institution = {Skillstore},
  year = {2026},
  number = {6},
  url = {https://skillstore.io/skills/0xbigboss-extract-transcripts/audits/6},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "extract-transcripts security audit report (audit version 6)"
version: "unspecified"
type: report
authors:
  - name: "0xBigBoss"
date-released: "2026-07-04"
url: "https://skillstore.io/skills/0xbigboss-extract-transcripts/audits/6"
identifiers:
  - type: other
    value: "skillstore:0xbigboss-extract-transcripts:audit:6"
    description: "Skillstore immutable audit report identifier"

Skillstore Score

Why this score Evidence Confidence: Medium

Architecture

Maintainability

Content

Community

Spec Compliance

What You Can Build

Review a Debugging Session

Convert a completed Claude Code session into markdown for later review and bug analysis.

Create Project Notes

Turn AI coding conversations into readable notes that explain decisions, commands, and outcomes.

Search Local Session History

Index past sessions and find conversations by keyword, project, or recent activity.

Try These Prompts

Extract One Session

Use extract-transcripts to convert this Claude Code session file into a readable markdown transcript: <session.jsonl>.

Summarize a Session Folder

Use extract-transcripts on this session directory, process all JSONL files, skip empty sessions, and return summary output only.

Export Detailed Transcript

Extract this session with tool calls and thinking blocks included, then save the markdown output to a private review file.

Index and Search Sessions

Build the DuckDB transcript index, search for "error handling", filter to my project, and show the most relevant session summary.

Best Practices

Review transcripts for secrets before sharing any output.
Use summary mode before exporting full conversations.
Store generated transcript files and indexes in private locations.

Avoid

Do not export all sessions into a shared folder without review.
Do not include thinking blocks unless they are required for analysis.
Do not index directories outside the intended session storage paths.

Frequently Asked Questions

What session formats does this skill support?

It supports Claude Code JSONL sessions, Codex CLI session JSONL files, and Codex history JSONL files.

Does it upload transcripts anywhere?

No network upload code was found. The scripts read and write local files only.

Can it process many sessions at once?

Yes. The Claude extractor can process all JSONL files in a directory, and the indexer can scan session folders.

Where does the searchable index live?

By default, the DuckDB index is stored under the local Claude transcript index directory.

Does it hide secrets in transcripts?

No. Users should review and redact secrets before sharing generated transcripts or index results.

What dependency is needed for indexing?

The indexer requires the duckdb Python package. Basic transcript extraction uses the Python standard library.

Developer Details

Author

0xBigBoss

License

MIT

Skillstore revision

Version notice

The author did not declare a version.

Repository

https://github.com/0xBigBoss/claude-code/tree/main/.claude/skills/extract-transcripts

Ref

b94843b1014e2052f8a7cd64f580a4023b48fd5f

Maintenance freshness

7/18/2026

Usage

53 downloads · 500 views

File structure

📄 extract_codex_transcript.py

📄 extract_transcript.py

📄 PLAN.md

📄 SKILL.md

📄 transcript_index.py