Continue the conversation — chat opens pre-seeded with the current signal, caps, and movement.
Agent Enablement Platform from Snyk founder Guy Podjarny — a package manager for AI agent skills with versioned installs, eval-scored quality, and (as of March 2026) Snyk-powered security scoring on every public skill.
Target user: enterprise platform teams and engineering organizations that have deployed multiple coding agents (Claude Code, Cursor, Gemini, Codex, OpenClaw, Copilot CLI, Copilot VSCode) and need consistent, vetted, governable skill distribution across them.
Differentiator: only platform combining package-manager discipline + automated evals + Snyk security scoring in one stack. Anthropic's official Claude Code marketplace is now built-in by default and 8 major skills marketplaces exist — but the three-part combination (versioning + evals + per-skill security audit) is unique to Tessl.
Adoption & Proof Points
- Named enterprise customers with attributed quotes, live on the Tessl homepage (May 2026):
- **Cisco** — John Groetzinger, Principal Engineer (Cisco skill measured at 1.79x improvement in real-world scenario evals)
- **HashiCorp/IBM** — Paul Thrasher, Director of Product, AI
- **ElevenLabs** — skills with review/scenario evals on the registry (prior '2x agent success' outcome)
- **PubNub** — suite of optimized skills with evals on the registry
- Homepage cites up to 3.3x agent improvement across 300+ libraries.
- Business signals: $125M total raised — structured as a $25M seed (April 2024) followed by a $100M Series A (Nov 2024, Index/Accel); not a single $125M Series A as previously described. No down-round, no layoffs detected; intact. Headcount ~58–60 (Crustdata 60 / +50% YoY; Tracxn 58 Feb 2026) — materially above the prior eval's '~21' undercount; 5 open roles. Registry indexes ~3,000 skills + version-matched docs for 10,000+ OSS packages; broader SKILL.md ecosystem grew ~450x in 14 weeks. CLI at v0.80.0 (native binary).
Recommended Use Cases
- **Enterprise skill governance at scale:** Platform teams standardizing agent behavior across 7+ agents (Claude Code, Cursor, Copilot, Gemini, etc.) — use Tessl's versioned installs, RBAC, org-wide skill distribution, and audit logs to enforce consistent, vetted skill sets without per-team drift.
- **Security-gated agent deployments:** Organizations where prompt injection, credential mishandling, or malware payloads in third-party skills are a blocking concern — Snyk-powered per-skill security scoring and CI-blocking on critically-flagged versions provide a defensible audit trail no other marketplace currently matches.
- **OSS library context injection:** Engineering teams building on 10,000+ indexed OSS packages who need agents to work with version-accurate documentation — `get_library_context` via MCP delivers version-matched docs at runtime, reducing hallucinated API usage.
- **Skill quality measurement and improvement:** Teams with existing SKILL.md files who want to quantify agent impact — task/scenario evals (with/without skill comparison) and the `skill-optimizer` (April 30, 2026) provide measured improvement baselines; Cisco's 1.79x outcome is the reference case.
- **CI/CD-integrated skill review:** Engineering orgs that want skill changes reviewed automatically on PRs — the `skill-review` GitHub Action and `setup-tessl` Action provide inline feedback without requiring developer context switches.
- **ISVs and platform vendors publishing skills:** Organizations like ElevenLabs and PubNub distributing agent skills to customers — the public registry with eval scoring and Snyk audits provides a credible, governed distribution channel versus raw GitHub publishing.
Risks & Limitations
- **No SOC 2 / ISO 27001 certifications:** blocks higher-confidence enterprise positioning; binding constraint for Assessed advancement
- **No independent third-party benchmarks or analyst coverage**
- **No Series B yet (~18 months since Series A):** runway and capital efficiency to watch
- **Competitive intensification:** Anthropic's official Claude Code marketplace ships built-in (with versioning + quality/security standards, though no built-in evals confirmed); 8+ skills marketplaces exist as of Q2 2026 (incl. Vercel-backed Skills.sh, Agensi) — structural distribution risk if Anthropic adds native evals
- **Thin Product Hunt / no G2 presence:** mainstream community-adoption signal is still light
- **Cloud-only:** no self-hosted/on-prem option; training-data opt-out unspecified
Capabilities & Integration
**Autonomy (6, Assistive): ** Tessl is infrastructure for agent skills, not an autonomous agent itself. `tessl init` auto-detects installed agents; `skill-optimizer` (April 30, 2026) measures and improves skill performance automatically. Skills enable other agents to be more autonomous — Tessl itself does not execute code, run multi-step workflows, or self-correct.
**Integration (7, Basic–Capable boundary):** Snyk integration adds CI-style security scanning to every skill; GitHub Action `setup-tessl` provides CI/CD integration; cross-agent compatibility verified across 7+ agents (Claude Code, Cursor, Gemini, Codex, OpenClaw, Copilot CLI, Copilot VSCode); `skill-review` GitHub Action automates PR review of SKILL.md files.
**Context (14, Repository-Deep):** Tessl's clearest moat. Registry indexes ~3,000 skills with eval-scored quality plus version-matched docs for 10,000+ OSS packages. Three eval methodologies — skill review (structural best-practice), task/scenario evals (runs the agent twice, with and without the skill, scoring the gap), and repo evals — plus the skill-optimizer that measures and auto-improves skills. MCP `get_library_context` ships version-matched documentation. Cross-repo, org-wide skills supported. Measured agent improvement 1.26x–3.3x (Cisco skill 1.79x).
**Interface (10, Dual-Mode): ** CLI (`@tessl/cli` at v0.80.0, now distributed as a native binary; npm path deprecated) + web registry (tessl.io/registry). `tessl init/install/search` work without login; `tessl doctor` diagnostics. MCP server enables agent integration across 7+ agents (claude-code, cursor, gemini, codex, openclaw, copilot, copilot-vscode). GitHub Action (`setup-tessl`) provides CI/CD entry. No dedicated IDE extension (MCP-mediated); no broad public API beyond CLI + MCP.