
tracking-pettracer-location
Retrieve and interpret an authorised owner's PetTracer collar inventory, latest known GPS fix, or bounded location history. Use for explicit PetTracer work, not generic pet questions or tracking a person. Distinguish measurement time, contact time, missing data, and stale fixes.
- Installs
- 96
- Last 24h
- —
- Trending rank
- —
- Security
- Not audited
Install
Works with every agent the skills CLI supports.
npx skills add https://github.com/tristanmanchester/agent-skills --skill tracking-pettracer-location- 01
Install
Run the command in your project. The CLI asks which agents to install for.
- 02
Describe the task
Ask your agent for something this skill covers. It loads tracking-pettracer-location automatically when relevant.
- 03
Review the result
The agent follows the SKILL.md steps, scripts and references bundled with the skill.
SKILL.md
The instructions your agent reads when the skill is triggered.
PetTracer location evidence
Keep this skill: it solves a specific owner's lookup task. No current official public developer API or replacement CLI was established during this review. The portal helper is therefore explicitly unofficial, not a vendor-supported client or a guarantee of live tracking. Prefer the official app/portal when its behaviour or the account contract differs.
Establish the target and permission
Use only an account/collar the user owns or is authorised to inspect. Confirm the actual device ID against the account inventory and official app before interpreting coordinates as the pet's location. A HomeStation's configured position is not a collar GPS fix. An absent device-type field remains unverified; do not infer type from an arbitrary name or the first item in a response.
Resolve SKILL_DIR to this installed skill directory. Set PETTRACER_TOKEN privately, or PETTRACER_USERNAME and PETTRACER_PASSWORD for an authorised login. Never ask for credentials in chat or pass them in command arguments. The helper does not store or print tokens and does not accept arbitrary API origins or credential redirects.
python "$SKILL_DIR/scripts/pettracer.py" list
python "$SKILL_DIR/scripts/pettracer.py" locate --device-id 12345 --max-age-seconds 900
Replace the sample ID with the verified collar ID. The 900-second default is a
local reporting threshold, not a manufacturer guarantee or the right threshold
for every tracking mode. recent means the supplied measurement timestamp passes
that chosen threshold; it does not mean the pet has stayed at those coordinates.
Read timestamps literally
Use only timeMeasure to compute fix age. Keep timeDb and lastContact separate;
new database activity/contact cannot make an old position fresh. Reject timezone-
less dates rather than silently calling them UTC. Unknown times and future times
are explicit states, never fresh by default. Display the actual fix time/timezone
and age alongside retrieval time.
The helper validates finite latitude/longitude ranges and preserves zero values.
It does not estimate a battery percentage or convert raw signal/accuracy fields
without a verified calibration/units contract. acc and horiPrec remain separate
raw quality fields with unverified units; do not call either a metre-radius merely
because a community integration did. home is a device flag, not proof the pet is
inside a precise geofence.
History
Use an explicit timezone-aware window, at most seven days per helper request:
python "$SKILL_DIR/scripts/pettracer.py" history --device-id 12345 \
--from-time 2026-09-13T06:00:00+02:00 --to-time 2026-09-13T12:00:00+02:00
These are example times, not the current requested interval. The helper preserves provider order and states completeness is unverified. Check returned timestamps against the requested interval; unknown/out-of-window points remain labelled. Sort or deduplicate deliberately for a route, preserving originals and gaps. Straight lines between sparse GPS fixes are not the actual path travelled.
Live updates and modes
For immediate owner search/live mode use the official app/portal. A subscription to updates does not itself command the collar to obtain a new fix. Changing modes can affect battery and device behaviour; do not change settings as a side effect of a location read. The old unbounded watcher is removed; retained protocol notes in portal and streaming describe the evidence needed for an explicitly scoped new integration, not a tested live client.
Do not promise continuous watching or future notification unless an authorised running service/scheduler has actually been established. A failed stream or no GPS fix is not evidence that the pet is safe, stationary, indoors, or at home.
Return an honest location result
State the selected collar, latest valid measured coordinates, measurement time, age/freshness state, and relevant missing/quality data. Report last contact separately. A map link can disclose precise location to a third party; include it only as appropriate to the user's request and keep raw history private. Do not overstate accuracy, reverse-geocode a precise address without need, or infer human routines.
Exit zero means the read was processed, not that a recent GPS fix exists. Inspect
last_fix.freshness and device_type_verified. Missing/invalid fixes, schema errors,
auth failures, and empty history are different outcomes. Per-request socket timeout
and a 2 MiB response bound are not a total wall-clock SLA. There are no automatic
retries or re-login loops; verify an unexpected endpoint/schema before retrying.
Maintainers: python -m unittest discover -s "$SKILL_DIR/tests" -v tests parsing,
selection, request boundaries, and mocked transport. No live account or collar was
queried in this review. Vendor operational help: https://help.pettracer.com/.
History results retain each original record and label its measurement time as inside,
outside, or unknown relative to the requested inclusive window. This is separate
from fix age. A non-object lastPos is a schema error, not evidence of no GPS fix.
FAQ
What does the tracking-pettracer-location skill do?
Retrieve and interpret an authorised owner's PetTracer collar inventory, latest known GPS fix, or bounded location history. Use for explicit PetTracer work, not generic pet questions or tracking a person. Distinguish measurement time, contact time, missing data, and stale fixes.
How do I install tracking-pettracer-location?
Run npx skills add https://github.com/tristanmanchester/agent-skills --skill tracking-pettracer-location in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/tracking-pettracer-location/ (personal) or .claude/skills/tracking-pettracer-location/ (project).
Which AI agents can use tracking-pettracer-location?
Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.
Is tracking-pettracer-location safe to install?
No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.
Who maintains tracking-pettracer-location?
It is published by tristanmanchester in tristanmanchester/agent-skills. It has 96 recorded installs.
More from tristanmanchester/agent-skills
designing-beautiful-websites
tristanmanchester/agent-skills
Design, implement, or critique website and web-app UX/UI: information structure, user flows, responsive layouts, visual systems, content, and interaction states. Use for a requested interface or visual-design change; do not expand a small component edit into an unrequested product redesign.
reddit-readonly
tristanmanchester/agent-skills
Browse and search Reddit in read-only mode using public JSON endpoints. Use when the user asks to browse subreddits, search for posts by topic, inspect comment threads, or build a shortlist of links to review and reply to manually.
animating-react-native-expo
tristanmanchester/agent-skills
Implement and debug React Native or Expo motion with Reanimated 4 and Gesture Handler 3: state transitions, layout, gestures, scroll, worklets, cancellation, and accessible alternatives. Use for native UI motion, not web Motion or a reason to add canvas graphics to ordinary views.
styling-nativewind-v4-expo
tristanmanchester/agent-skills
Set up, repair, and build components with NativeWind v4 and Tailwind CSS v3 in Expo React Native apps. Use for an explicitly v4 project, v4 configuration failures, className interoperability, or v4 theming; not for NativeWind v5 or generic web Tailwind setup.
nextjs-framer-motion-animations
tristanmanchester/agent-skills
Implement and debug Motion for React in Next.js: local interaction, presence, layout, shared elements, scroll, reduced motion, and router integration. Use for explicit Motion/Framer Motion work or interactions that need it; prefer CSS for simple effects and framework-native view transitions for supported route cases.
react-native-skia
tristanmanchester/agent-skills
Build, debug, or profile React Native Skia canvases, paths, rich text, shaders, images, snapshots, video, and animated graphics. Use when Skia is selected or custom drawing is needed. Do not introduce a canvas merely because ordinary native UI should look polished or animate smoothly.
Popular ai & agents skills
google-agents-cli-publish
google/agents-cli
This skill should be used when the user wants to "publish an agent", "publish my ADK agent", "register an agent with Gemini", "publish to Gemini" (formerly Gemini Enterprise), or needs guidance on the agents-cli publish gemini command. Also use when the user wants to "manage agents in Agent Registry", "list/update/delete registered agents", or "register an MCP server". Covers ADK vs A2A registration modes, programmatic and interactive usage, flag reference, auto-detection from deployment metadata, Agent Registry fleet management, and troubleshooting. Part of the agents-cli skills suite. Do NOT use for deployment (use google-agents-cli-deploy).
google-agents-cli-eval
google/agents-cli
This skill should be used when the user wants to "run an evaluation", "evaluate my agent", "evaluate my ADK agent", "write an eval dataset", "analyze eval failures", "compare eval results", "optimize agent", or needs guidance on the Gemini platform eval methodology and the Quality Flywheel. Covers eval metrics, dataset schema, LLM-as-judge scoring, and common failure causes. Applies to any agents-cli project, whatever framework the agent is written in. Do NOT use for agent API code patterns (ADK: use google-agents-cli-adk-code), deployment (use google-agents-cli-deploy), or project scaffolding (use google-agents-cli-scaffold).
skill-creator
anthropics/skills
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
codex-pet
prime-skills/runcomfy-agent-skills
Codex Pet generator on RunComfy. Build a Codex-compatible Codex Pet spritesheet.webp + pet.json from a single reference image, drop it into ${CODEXHOME:-$HOME/.codex}/pets/ / and Codex picks it up as a custom Codex Pet next to the 8 built-ins. This skill produces the exact Codex Pet atlas Codex expects (1536x1872 PNG/WebP, 8 cols x 9 rows, 192x208 cells, 9 animation states — idle, running-right, running-left, waving, jumping, failed, waiting, running, review). Calls OpenAI GPT Image 2 edit ONCE via the local RunComfy CLI as runcomfy run openai/gpt-image-2/edit to produce a canonical Codex Pet pose, then assembles all 9 animation rows programmatically with ImageMagick micro-transforms — no Codex Pro, no $imagegen, no OPENAIAPIKEY required, only RUNCOMFYTOKEN. Triggers on "codex pet", "create codex pet", "make codex pet", "hatch codex pet", "/hatch image", "desktop pet codex", "codex pets", "spritesheet.webp", or any explicit ask to build a custom pet for OpenAI Codex.
full-output-enforcement
leonxlnx/taste-skill
Overrides default LLM truncation behavior. Enforces complete code generation, bans placeholder patterns, and handles token-limit splits cleanly. Apply to any task requiring exhaustive, unabridged output.
executing-plans
obra/superpowers
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available