How do I install the Overcast Where skill?

Run npx skillmds@latest add kdr/overcast-where in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Overcast Where safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Overcast Where?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Overcast Where free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Overcast Where?

kdr (@kdr) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Overcast Where by kdr — Agent Skill

Name: Overcast Where
Author: kdr

File contents

overcast-where

Use this skill to turn a moment into spatial evidence: a bounding box on the target plus a verified crop. It follows the detector-proposes / VLM-verifies pattern (open-vocab detection like OWLv2, then confirm the crop) — because chat VLMs are unreliable at emitting raw coordinates, so a real detector draws the box and the VLM only judges the crop. Use the broad overcast skill and overcast/reference/verbs.md for exact flags.

Setup

see --detect needs a detection provider bound (boxes come from OWLv2, not the brain LLM):

scripts/visual-db-uv.sh --detect     # once: prints DETECT_PY (the venv python)
export DETECT_PY="$DETECT_PY"; overcast provider setup apply --preset owl-local --yes --json  # persists a portable shipped: ref for detect.py + the venv python

Workflow

Have the moment (use overcast-pinpoint / overcast-frame-grid): timestamp T on record REC.
Detect the target in that frame, then materialize + verify the box:

overcast see frame://REC@T --detect "<target phrase>" --json      # -> see record with detections[]
overcast crop <see-record-id> --all --class "<target phrase>" --pad 0.15 --json
overcast see <crop-path> --prompt "Does this crop show <target>? yes/no + describe" --json

The re-see of each crop is what kills false positives — open-vocab detectors emit confident boxes for almost any phrase at low thresholds.

Optionally sharpen the exhibit and record the finding:

overcast enhance <crop-path> --ops upscale,denoise --json
overcast finding create "<target> located at T" --ref <see-record-id> --confidence medium --json
overcast brief --export ./where.md --json

Output

Per confirmed target: the timestamp, the box (from the see --detect record), the crop path, and the verification verdict. Cite the see detection record.id + media.at; note the crop is the durable, memory-friendly evidence artifact.

Caveats

Never ask the brain LLM for coordinates directly — bind a detector and verify crops. Detector confidence is not calibrated across free-form phrases: a high score on a rare phrase can still be wrong, so the crop re-check decides. For a specific PERSON, use face --match instead of --detect. Boxes are per sampled frame, not tracks.

---name: overcast-wheredescription: Locate WHERE in a frame a target is — open-vocab detect, crop the box, then VLM-verify the crop so hallucinated boxes don't survive.---# overcast-whereUse this skill to turn a moment into spatial evidence: a bounding box on thetarget plus a verified crop. It follows the detector-proposes / VLM-verifiespattern (open-vocab detection like OWLv2, then confirm the crop) — because chatVLMs are unreliable at emitting raw coordinates, so a real detector draws the boxand the VLM only judges the crop. Use the broad `overcast` skill and`overcast/reference/verbs.md` for exact flags.## Setup`see --detect` needs a detection provider bound (boxes come from OWLv2, not thebrain LLM):```bashscripts/visual-db-uv.sh --detect # once: prints DETECT_PY (the venv python)export DETECT_PY="$DETECT_PY"; overcast provider setup apply --preset owl-local --yes --json # persists a portable shipped: ref for detect.py + the venv python```## Workflow1. Have the moment (use `overcast-pinpoint` / `overcast-frame-grid`): timestamp T on record REC.2. Detect the target in that frame, then materialize + verify the box:```bashovercast see frame://REC@T --detect "<target phrase>" --json # -> see record with detections[]overcast crop <see-record-id> --all --class "<target phrase>" --pad 0.15 --jsonovercast see <crop-path> --prompt "Does this crop show <target>? yes/no + describe" --json``` The re-`see` of each crop is what kills false positives — open-vocab detectors emit confident boxes for almost any phrase at low thresholds.3. Optionally sharpen the exhibit and record the finding:```bashovercast enhance <crop-path> --ops upscale,denoise --jsonovercast finding create "<target> located at T" --ref <see-record-id> --confidence medium --jsonovercast brief --export ./where.md --json```## OutputPer confirmed target: the timestamp, the box (from the `see --detect` record),the crop path, and the verification verdict. Cite the `see` detection`record.id` + `media.at`; note the crop is the durable, memory-friendly evidenceartifact.## CaveatsNever ask the brain LLM for coordinates directly — bind a detector and verifycrops. Detector confidence is not calibrated across free-form phrases: a highscore on a rare phrase can still be wrong, so the crop re-check decides. For aspecific PERSON, use `face --match` instead of `--detect`. Boxes are per sampledframe, not tracks.

Frequently asked questions

Overcast Where