Platform API fallback CLI for unsupported tasks stream-first

[anand-testcompare]

Why This Exists

palantir-mcp does not cover every operation needed for real delivery. Compute-module workflows (especially stream-related setup/validation tasks) need a reliable fallback path.

Desired Outcome

A token-efficient CLI for high-value unsupported tasks so agents can finish workflows without fragile UI click-ops.

Scope

In scope:

• Platform API fallback CLI using existing env auth (FOUNDRY_URL, FOUNDRY_TOKEN).
• Stream-first task surface for compute-module workflows.
• Concise agent-friendly output (--json plus terse human mode).
• Clear, actionable errors.

Out of scope:

• Replacing palantir-mcp for common operations.
• Full Foundry API parity in v1.
• Browser/HAR automation.

Success Criteria

• Auth works with existing env vars and no secret persistence.
• Stream-first fallback tasks are executable and documented.
• Output is stable, concise, and parseable by agents.
• Error messages include one clear next step.

High-Level Plan

1. Lock v1 command surface around stream-first compute tasks.
2. Implement API client bootstrap + command routing.
3. Add concise output modes and error normalization.
4. Add integration docs for agent workflows.

Dependencies

• Blocked by: Compute module playbook decision guide constraints and references #23
• Related: Foundry defaults: hard-signature profiles and unified default policy #22, Auto-download Palantir docs database on plugin install (post-install hook) #19

[shpit-agent]

https://github.com/steipete/mcporter

lol maybe i dont need to roll that cli myself. this was almost exactly what i wanted.. honestly maybe a good idea to replace various of our current mcps @anand-testcompare