praktor: Telegram upload traversal permits persistent agent instruction injection
- Last modified
- Where the record stands: published, rejected by its publisher, or superseded by another id.What do these mean? →
- published
- Whether CISA lists it in its Known Exploited Vulnerabilities catalog: proof that attackers are using it.What do these mean? →
- Not listed in CISA KEV
- How many real-world setups can hit the bug, from the vendor's advisory and CISA data. It does not know your setup.What do these mean? →
- Default setups
- Fix
- Fixed in 0.9.1
- Whether working attack code is public (Exploit-DB, Metasploit, Nuclei or GitHub). Public code makes attacks easier.What do these mean? →
- None on Exploit-DB, Metasploit or GitHub
- Exploitation: none
- Automatable: yes
- Technical impact: partial
Summary
The praktor multi-agent Claude Code orchestrator (its Telegram bot component) is affected, and in its default configuration any Telegram user can exploit it by uploading a document with a crafted filename. This lets an attacker write files into the persistent agent workspace, such as CLAUDE.md or configuration and hook files, which the agent then reads in later runs, allowing their malicious instructions to persist.
Summarized by anthropic/claude-haiku-5.5. Verify details against the original sources.
Read the original description
praktor is a multi-agent Claude Code orchestrator with Telegram input and output, Docker isolation, and a Mission Control UI. Prior to 0.9.1, both Telegram upload sites in `internal/telegram/bot.go` concatenate the attacker-controlled document filename into a destination path without sanitization, and `path.Join` collapses parent-directory traversal. A remote Telegram user who can message the bot can therefore write outside `uploads/` within the persistent agent workspace mounted at `/workspace/agent`. Because `allow_from: []` permits all senders by default, no prior authentication or privilege is required in that configuration. The attacker can plant or overwrite `CLAUDE.md`, instruction, skill, hook, or configuration files that the Claude agent reads and operates on across later runs, turning a document upload into stored instruction injection. The demonstrated arbitrary-write scope is the persistent agent workspace volume, not an unrestricted host-filesystem write. This issue is fixed in version 0.9.1.
Timeline
- CVE published
- First seen by cve.now
- CISA SSVC assessment
Exploitation: none · Automatable: yes · Technical impact: partial
- Record last updated
Affected versions
- mtzanidakis praktor
fixed in 0.9.1
References 4
GitHub · mtzanidakis/praktor
Questions about CVE-2026-62331
What does this mean?
A filename sent with a Telegram document upload is joined into the save path without cleaning, so it can point outside the uploads folder. An attacker can then write files into the agent's persistent workspace, such as instruction files like CLAUDE.md, which the agent later reads.Source: github.com
Am I affected, and how do I check?
Versions v0.9.0 and earlier of praktor, along with the development HEAD at commit 3902d29, are affected. The setup matters: the bot must accept document uploads and the allow_from setting must permit all senders, which is the default. Check the installed version and the telegram.allow_from value in the configuration.Source: github.com
What should I do about it?
Upgrade praktor to v0.9.1, which fixes the problem and is the recommended fix. If upgrading is not possible right away, restrict telegram.allow_from to trusted user IDs.Source: github.com
Is it being exploited?
It is not in CISA's Known Exploited Vulnerabilities catalog, and CISA's SSVC assessment does not mark exploitation as active. That does not mean it has not been attacked: vendors and researchers can report attacks before CISA lists them. No public exploit code is tracked.Source: cve.now feed data